What is an unconscious representation?

In contemporary cognitive neuroscience, we often encounter the term “unconscious representations” (e.g., Shea & Frith, 2016). These are not just representations that could be conscious but just so happen to be unconscious; they are a distinct type of cognitive representation, where their status as unconscious is inherent to the kind of representations they happen to be. That is, being unconscious is part of the representational format in question.

This conception of unconsciousness differs from a type of established folk model inherited from Freud’s early twentieth-century armchair psychology. Here, the unconscious is just a place (e.g., “in the mind”) where representations go (by implication, consciousness is also a mind place, just a place different from the unconscious place). According to this picture, representations are unconscious as long as they are in that place (e.g., the unconscious), usually banished there by some speculative mental process (itself unconscious!) called “repression.” Because being unconscious is a contingent and not inherent aspect of those representations, they could lose it by being “brought back” to consciousness via some therapeutic intervention. This type of “folk Freudianism” is a deleterious brain cramp that needs to be abandoned if the aim is to understand how contemporary cognitive neuroscientists understand unconscious representations. Even when cleansed of its Freudian associations, as in more recent talk of the “cognitive unconscious” (Reber, 1993) or the “unconscious mind” (Bargh & Morsella, 2008), the idea of the unconscious as a “place” (or “brain system”) where representations reside is misleading and should be jettisoned. Instead, we should speak of unconsciousness as an inherent property of some representations in the brain.

Let’s reiterate. The unconscious is not a place where previously conscious representations go. Instead, it is a fundamental property of the vehicles in which some representations (crucial for various cognitive functions) are instantiated in the brain. These representations are inherently unconscious. They cannot be “brought” “into” consciousness (a seductive remnant of the unconscious-as-place conceptual metaphor) by any conceivable procedure. However, cognitive scientists and linguists can generate public representations (e.g., instantiated in some kind of linguistic theory or neuroscientific computational model) that redescribe these unconscious representations to establish their content (e.g., what they are designed to represent) and function (their role in the cognitive economy of the agent). These public representations, clearly phenomenally accessible to all as conscious agents, are convenient representational redescriptions (RRs) of the content of what are inherently unconscious representations.

For instance, psycholinguists sometimes try to reconstruct the underlying unconscious representations we use to parse the linguistic input’s syntactic and phonetic structure during language comprehension by transforming them (or, more accurately, their best guesses as to which these are) into all forms of public representations, like the ubiquitous sentence tree diagrams of generative linguistics or the pictorial diagrams of cognitive grammar (Jackendoff, 1987; Langacker, 2008). These public representations are not intended to be the exact or literal analogs of the unconscious representations people employ to parse the syntactic structure of a sentence. For one, they are in a different format (digital or paper and pencil diagrams) distinct from the target unconscious representations, which exist as (perhaps structurally similar) activation and connection patterns in neuronal assemblies. However, public representations of unconscious representation used in cognitive-scientific theorizing are designed to preserve the representational content of the underlying unconscious representations (the what of what is being represented) across distinct vehicles. So, both a sentence tree grammar and the underlying unconscious representations that enable us to parse the syntactic structure of a sentence represent the same content (e.g., the target sentence’s synaptic structure).

If unconscious representations are inherently unconscious, how do we even know they exist? Perhaps all the diagrams produced by linguists and cognitive scientists are just made up, referring to nothing since we cannot observe unconscious representations. There are two responses to this worry.

First, the idea of unconscious neural representations as “unobservable posits” is greatly exaggerated because unobservability (unlike unconsciousness!) happens to be not an inherent but merely a contingent property of unconscious brain representations. This is because the unobservability of any entity, including unconscious representations, is a two-place relational property, always relative to our historically fluid capacities (and limitations) as scientific observers. This means that just like in other scientific fields like high-energy physics, molecular biology, or astronomy, previously unobservable entities can cross the threshold of observability after some kind of technological advancement occurs in our observational instruments. Some contemporary cognitive neuroscientists argue that this is precisely what has happened to unconscious brain representations, which are now, given advances in fMRI technology, as observable as apples, tables, and chairs (Thomson & Piccinini, 2018).

Second, even if we treat unconscious representations as classic unobservables, we still can still have a solid warrant for their existence based on their (presumably explanatorily successful) role in our best cognitive theories and models. For instance, the only way to rationally reconstruct how cognitive neuroscientists proceed as scientists is thus via our old friend abduction (inference to the best explanation). Like every other scientist, cognitive neuroscientists proceed from observing an initially puzzling phenomenon to trying to understand the generative mechanisms that produce that phenomenon and, thus, solve the puzzle.

In the case of unconscious representations, the observed phenomenon is usually some kind of initially puzzling human ability or capacity no one doubts exists—like the ability to parse the phonetic structure of words or the syntactic structure of sentences or engage in fine-grained motor control. Unconscious representations then come in as proper parts of the underlying cognitive mechanisms posited by the neuroscientific theory, whose operations account for the phenomenon in question, thus solving the puzzle of how people “can do that” (Craver, 1998). Cognitive scientists thus posit the existence of unconscious representations as part of this underlying mechanism because they provide the best explanation, thus accounting for the puzzle of people’s ability to exercise the target capacity. Inference to the best explanation thus justifies both the act of the positing and the reality of the (for the sake of argument “unobservable”) unconscious representations featuring in our most explanatory successful models of how people can exercise a given capacity (Boyd, 1983).

To sum up, unconscious representations, like those accounting for your capacity to parse the phonetic structure of every word in this paragraph as you read it, are a completely uncontroversial part of the scientific ontology of contemporary cognitive neuroscience. They feature centrally in almost every mechanistic model of every cognitive capacity and ability we have, providing a solid scientific account of some of the essential functions of the mind. Their status as “unobservable” has been overstated since now we have routine access to them as observable entities and even the processes in which they participate.

References

Bargh, J. A., & Morsella, E. (2008). The Unconscious Mind. Perspectives on Psychological Science: A Journal of the Association for Psychological Science, 3(1), 73–79.

Boyd, R. N. (1983). On the Current Status of the Issue of Scientific Realism. In C. G. Hempel, H. Putnam, & W. K. Essler (Eds.), Methodology, Epistemology, and Philosophy of Science: Essays in Honour of Wolfgang Stegmüller on the Occasion of His 60th Birthday, June 3rd, 1983 (pp. 45–90). Springer Netherlands.

Craver, C. F. (1998). Neural Mechanisms: On the Structure, Function, and Development of Theories in Neurobiology [University of Pittsburgh]. https://philpapers.org/rec/CRANMO-2

Reber, A. S. (1993). Implicit Learning and Tacit Knowledge: An Essay on the Cognitive Unconscious. Oxford University Press.

Shea, N., & Frith, C. D. (2016). Dual-process theories and consciousness: the case for “Type Zero” cognition. Neuroscience of Consciousness, 2016(1), niw005.

Thomson, E., & Piccinini, G. (2018). Neural representations observed. Minds and Machines, 28(1), 191–235.

Are We Cognitively Susceptible to Tests?

In one the clearest statements about the difference it makes to emphasize cognition in the study of culture and, more generally, for the social sciences as a whole, the anthropologist Maurice Bloch (2012) writes that, if we consider closely every time we use the word “meaning” in social science, then “a moment’s reflection will reveal that ‘meaning’ can only signify ‘meaning for people’. To talk of, for example, ‘the meaning of cultural symbols’, as though this could be separated from what these symbols mean, for one or a number of individuals, can never be legitimate. This being so, an absolute distinction between public symbols and private thought becomes unsustainable” (4).

As a critique of Geertzian and neo-Diltheyan arguments for “public meaning” and “cultural order” sui generis, Bloch’s point is fundamental, as it reveals a core problem with arguments built on those foundations once they have been untethered from “meaning for people” and can almost entirely be given over to “meaning for analysts.” Yet, and as Bloch makes it a point to emphasize, such critiques can only get us so far in attempting to change practices, as even if “a moment’s reflection” like this may lead some to agree with Bloch’s claim, without an alternative, these models will persist more or less unchanged. If “meaning for people” stands as some equivalent for a tethering to cognitive science as recommended by theorists like Stephen Turner (2007), then what is needed is a programmatic way of doing social theory without “minimizing the cognitive” by attempting, instead, to bridge social theory and cognitive neuroscience in the design of concepts.

In fairness to Geertz, one of his more overlooked essays proposes a culture concept that seems to want to avoid the very problem that Bloch identifies. In “The Growth of Culture and the Evolution of Mind” Geertz (1973) draws a connection between culture and “man’s nervous system,” emphasizing in particular the interaction of culture and the (evolved) mind in the following terms: “Like a frightened animal, a frightened man may run, hide, bluster, dissemble, placate or, desperate with panic, attack; but in his case the precise patterning of such overt acts is guided predominantly by cultural rather than genetic templates.” Here the problem of relating the cultural to the cognitive seems clearly resolved, as the latter is reduced to “genetic templates.” Yet, contrary to Sewell’s (2005) positive estimation of this aspect of Geertz’s thought as “materialist,” we should be wary of taking lessons from Geertz if by “materialist” Sewell means a culture concept that does due diligence to the evolved, embodied, and finite organisms we all are. Nonetheless, in many respects, the Geertzian move still prevails in contemporary cultural sociology which, likewise, features an admission of the relevance of the cognitive to the cultural, but retains a similar bracketing as de facto for figuring out the thorny culture + cognition relation.

For instance, recently Mast (2020) has emphasized that “representation” (qua the proverbial turtle) works all the way down, even in the most neurocognitive of dimensions, and so we cannot jettison culture even if we want to include a focus on cognition because we need cultural theory to account for representation. Likewise, Norton (2018) makes a similar claim by drawing a distributed cognition framework into sociology, but making “semiotics” the ingredient for which we need a designated form of cultural theory (in this case, his take on Peircean “semeiotics”) to understand. Kurkian (2020), meanwhile, argues that unless we admit distinguishably cultural ingredients like these, attempting any sort of marriage of culture + cognition will fail, because cognition will be about something that does not tread on culture’s terrain, like “information” for instance.

Each of these is a worthwhile effort, yet in some manner they misunderstand the task at hand in attempting a culture + cognition framework, recapitulating what Geertz did in 1973. This is because any such framework must rest on new concept-formation rather than what amounts to a defense of established concepts. This would admit that cultural theories of the past cannot be so straightforwardly repurposed without amendments. What we tend to see, rather, are associations of culture concepts (semiotics, representation) and cognitive concepts (distributed cognition, mirror neurons) by drawing essentially arbitrary analogies and parallels between concepts that otherwise remain unchanged. In most cases, such a bracketed application replicates the disciplinary division of labor in thought because the onus is never placed on revision, despite the dialectical encounter and the possibilities that each bank of concepts presents to the deficiencies and arbitrariness of the other. We either hold firm to our cultural theories of choice, or we engage in elaborate mimicry of a STEM-like distant relation.

Following Deleuze (1995), we should appreciate that to “form concepts” is at the very least “to do something,” like, for instance, making it wrong to answer the question “what is justice?” by pointing out a particular instance of justice that happened to me last weekend. Deleuze adds insight in saying that concepts attempt to find “singularities” from within a “continuous flow.” The insight is apt to the degree that culture + cognition thinking seems rooted in the sense that there is a “flow” here and that, maybe, the concepts we’ve inherited, most of them formed over the last 80 years, that make culture and cognition “singular” are simply not helpful anymore. Yet to rehash settled, unrevised cultural theories and bring them into relation with emerging cognitive theories (also unchanged) is essentially to “do” something with our concepts like affirm a thick boundary between sociologists’ jurisdiction and cognitive science’s jurisdiction, forbidding anything that looks like culture + cognition, and, in all likelihood, creating only an awkward, fraught, short-lived marriage between the two, which, despite the best of intentions, will continue to “minimize mentalistic content,” have the effect of carefully limiting the role that “psychologically realistic mechanisms” can play in concept-formation, and which will, in retrospect, probably only produce a brand of social theory that will seem hopelessly antique for sociologists looking back from the vantage of a future state of the field, one possibly even more removed from present-time concerns with “cognitive entanglements.”

The task should instead be something akin to what Bourdieu (1991) once called “dual reference” in his attempt to account for the strange verbiage littered throughout Heidegger’s philosophy (dasein, Sorge, etc). For Bourdieu, Heidegger’s work remains incomprehensible to us if we reference only the philosophical field in which he worked, and likewise incomprehensible if we reference only the Weimar-era political field in which he was firmly implanted. Instead, Heidegger’s philosophy, in particular these keywords, consists of position-takings in both fields simultaneously, which for Bourdieu goes some way in explaining the strange and tortured reception of Heidegger (with Being and Time something of a bestseller in Germany when published in 1927 and still canonical in pop philosophy pursuits today) to present-day.

Thus, in forming concepts, the goal should not be to posit an order of influence (culture → cognition, cognition → culture), nor to bracket the two (culture / cognition) and state triumphantly that this is where culture concepts can be brought to bear and this where cognitive ones can be, leaving both unchanged. Norton is right: Peirce has lots of bearing on contemporary cognitive science (see Menary 2015). But to say this and not amend an understanding of semeiotics (which, it seems, Peirce would probably advocate were he alive today, as he always considered his semeiotics as a branch of the “natural science” he always pursued) is a non-starter.

My argument is that concept-formation of the culture + cognition kind should yield dual reference concepts rather than bracketing concepts or order of influence concepts. The proposal will be that the concept of “test” demonstrates such a dual reference concept. We cannot account for the apparent ubiquity of tests, why they are meaningful, and how they are meaningful without reference to both a cognitive mechanism and a sociohistorical configuration that combines with, appropriates, and evokes it. The analysis here involves genealogy, institutional practice, site-specificity, and social relations.

Elsewhere (Strand 2020) I have advocated a culture + cognition styled approach as the production of “extraordinary discourse” and, relatedly, as concept-formation that can be adequate for “empirical cognition” as a neglected, minor tradition since the time of Kant (Strand 2021; though one with a healthy presence in classical theory). More recently, Omar and I have attempted concept-formation that more or less looks like this in recommending a probabilistic revision of basic tenets of the theory of action (forthcoming, forthcoming). To put it starkly: we need new concepts if we want something like culture + cognition. To work under the heading of “cognitive social science” is akin to a compass-like designation in a new direction. And rather like Omar (2014) has said, if theorists, so often these days casting about for a new conversation to be part of now that “cultural theory” is largely exhausted and we can only play with the pieces, want a model for this kind of work, they might study the role that philosophers have come to play in cognitive science, as engaged in what very much seems like a project of concept-formation.

In this post, I will attempt something similar, more generally as a version of deciphering “meaning for people” by asking a simple question: Why are tests so meaningful and seemingly ubiquitous in social life (Marres and Stark 2020; Ronnell 2005; Pinch 1993; Potthast 2017)? I will consider a potential “susceptibility” to tests and why this might explain why we find them featured so fundamentally in areas as varied as education, science, interpersonal relationships, medicine, morality, technology, and religion, as a short list, and how they can be given a truly generalized significance if we conceptualize test as trial (Latour 1988). More generally, the new(ish) “French pragmatist sociology” has made the epreuve (what mutually translates “test” and “trial” into French) a core concept as a way of “appreciating the endemic uncertainty of social life” (Lemieux 2008) though without implying too much about what a cognitive-heavy phase like “endemic uncertainty” might mean. The French pragmatists [1] might be on to something: test or trial may qualify as a “total social phenomena” in the tradition of Mauss (1966), less because we can single out one test as “at once a religious, economic, political, family, phenomena” and more because each of these orders depends, in some manner, on tests. This is more fitting with a cognitive susceptibility perspective, as I will articulate further below.

Provisionally, I will define a test as the creation of uncertainty, a suspension of possibilities, a way of “inviting chance in,” for the purpose of then resettling those possibilities and resolving that uncertainty by singling out a specific performance. After a duration of time has elapsed, the performance is complete. The state of affairs found at the end is what we can call an “outcome,” and it carries a certain kind of “objective” status to the extent that the initial uncertainty or open possibility is different now, less apparent than it was before, and “final” in some distinguishable way.

If testing appears ubiquitous and “total,” this is not because tests necessarily work better than other potential alternatives as ways of handling “endemic uncertainty.” It is also not because testing features as part of some larger cultural process in motion (like “modernity’s fascination with breaking known limitations” [Ronnell 2005]). Rather, I want to claim that if tests are ubiquitous, this indicates a cognitive susceptibility to tests, thus revealing latent “dispositions,” such that we could not help but find tests “meaningful for people” like us. Some potential reasons why are suggested by referencing a basic predictive processing mechanism:

According to [predictive processing], brains do not sit back and receive information from the world, form truth evaluable representations of it, and only then work out and implement action plans. Instead brains, tirelessly and proactively, are forever trying to look ahead in order to ensure that we have an adequate practical grip on the world in the here and now. Focused primarily on action and intervention, their basic work is to make the best possible predictions about what the world is throwing at us. The job of brains is to aid the organisms they inhabit, in ways that are sensitive to the regularities of the situations organisms inhabit (Hutto 2018).

Thus, in this rendering, we cannot help but notice “sensory perturbations” as those elements of our sensory profile that defy our expectation (or, in more “contentful” terms, our predictions). These errors stand out as what we perceive, and we attend to them by either adjusting ourselves to fit with the error (like sitting up a little more comfortably in our chair) or by acting to change those errors, so that we do not notice them anymore. In basic terms, then, the predictive processing “disposition” involves an enactive engagement with the world that seeks some circumstance in which nothing is perceived, because, we might say, everything is “meaningful” (i.e. expected). If we define “meaning” as something akin to “whatever subjectively defined qualities of one’s life make active persistence appealing,” then this adaptation of the test concept might be a way of accounting for meaning without a “minimum of mentalistic content” while incorporating a “psychologically realistic mechanism” (Turner 2007).

In what follows I will examine whether there is some alignment between this disposition and tests as a ubiquitous social process. If so, then it may be worthwhile to build on the foundation laid by the French pragmatists for concept-formation of the culture + cognition kind.

On cognitive susceptibility

The notion of cognitive “susceptibility” is drawn from Dan Sperber (1985) and the idea that, rather than dispositions that create a more direct link between cognition and cultural forms, that link may more frequently operate as susceptibility.

Dispositions have been positively selected in the process of biological evolution; susceptibilities are side-effects of dispositions. Susceptibilities which have strong adverse effects on adaptation get eliminated with the susceptible organisms. Susceptibilities which have strong positive effects may, over time, be positively selected and become, therefore, indistinguishable from dispositions. Most susceptibilities, though, have only marginal effects on adaptation; they owe their existence to the selective pressure that has weighed, not on them, but on the disposition of which they are a side-effect (80-81).

Sperber uses the example of religion. “Meta-representation” is an evolved cognitive disposition to create mental representations that do not have to pass the rigorous tests that apply to everyday knowledge. It enables representations not just of environmental and somatic phenomena, but even of “information that is not fully understood” (83). Because it has these capabilities, the meta-representational disposition creates “remarkable susceptibilities. The obvious function served by the ability to entertain half-understood concepts and ideas is to provide intermediate steps towards their full understanding. It also creates, however, the possibility for conceptual mysteries, which no amount of processing could ever clarify, to invade human minds” (84). Thus, Sperber concludes that “unlike everyday empirical knowledge, religious beliefs develop not because of a disposition, but because of a susceptibility” (85).

The disposition/susceptibility distinction can be quite helpful in navigating the murky waters around Bloch’s trope of “meaning for people,” because we do not necessarily have to give cultural forms over directly to dispositions. Rather, those cultural forms can arise as susceptibilities, which offer far more bandwidth to capture the cognitive dimensions of cultural forms as instances of “meaning for people.”

Thus, when God “tests the faith” of Abraham by ordering him to sacrifice his child Isaac, a space of chances is opened, and depending on how the test goes, something about Abraham will become definitive, at least for a while. A perceived lack of faith becomes equivalent to a noticeable error here, and it can be resolved by absorbing this uncertainty through some process that generates an outcome to that effect. Even though Abraham does not end up sacrificing Isaac in the story, he was prepared to do so, and thus he “proves” his faith. Some equivalent to this “sacrifice” remains integral to tests of faith of all sorts (Daly 1977).

I hypothesize that there must be a (cognitive) reason why this test, and the whole host of others we might come across, in fields and pursuits far removed from Abrahamic religion, is found in moments like these and in situations that mimic (even vaguely) God’s “test” of Abraham. The role of tests in this religious tradition, and potentially as a total social phenomenon, indicates something about “susceptibility” (in Sperber’s sense) to them. “Disposition” in this case concerns the predictive processing disposition to eliminate prediction error by either adapting a generative model to the error or by acting to change the source of the error; either way, our expectations change and we do not notice what stood out for us before. For tests, the construction of uncertainty and more possibilities than will ultimately be realized is a kind of susceptibility that corresponds to the predictive disposition. More specifically, this means that tests allow something to be known to us by enabling us to expect things of it.

Tests: scientific, technological, moral

What is remarkable about this is the range of circumstances to which we turn to tests to construct our expectations. Consider Latour’s description of the Pasteur’s experimental technique:

How does Pasteur’s own account of the first drama of his text modify the common sense understanding of fabrication? Let us say that in his laboratory in Lille Pasteur is designing and actor. How does he do this? One now traditional way to account for this feat is to say that Pasteur designs trials for the actor to show its mettle. Why is an actor defined through trials? Because there is no other way to define an actor but through its actions, and there is no other way to define an action but by asking what other actors are modified, transformed, perturbed or created by the character that is the focus of attention … Something else is necessary to grant an x an essence, to make it into an actor: the series of laboratory trials through which the object x proves it mettle … We do not know what it is, but we know what it does from the trials conducted in the lab. A series of performances precedes the definition of the competence that will later be made the sole cause of these performances (1999: 122, 119).

Here the test (or “trial”) design works in an experimental fashion by exposing a given yeast ferment to different substances, under various conditions just to see what it would do. By figuring this out, Pasteur “designs an actor,” which we can rephrase as knowing an object by now being able to hold expectations of it, being able to make predictions about it, and therefore no longer needing to fear what it might do or even have to notice it.

Latour is far from alone in putting such emphasis on testing for the purposes of science. Karl Popper (1997), for instance, insists on the centrality of the test and its trial function: “Instead of discussing the ‘probability’ of a hypothesis we should try to assess what tests, what trials, it has withstood; that is, we should try to assess how far it has been able to prove its fitness to survive by standing up to tests. In brief, we should try to assess how far it has been ‘corroborated.’” To put a hypothesis on trial is, then, to imperil its existence, as an act of humility. Furthermore, it is to relinquish one’s own claim over the hypothesis. If a “test of survival” is the metric of scientific worth, then one scientist cannot single-handedly claim control: hypotheses need “corroboration,” a word which Popper prefers over “confirmation” because corroboration suggests something collective.

When Popper delineates the nuances of the scientific test, he also seems to establish tests for membership in a scientific community, as based on this sort of collective orientation, which requires individual humility, and in which, from the individual scientist’s standpoint means “inviting chance in” relative to their own hypothesis, making them subject to more possibilities than what the scientist might individually intend, including the possibility that they could be completely wrong.

Meanwhile, in Pinch’s approach, which focuses specifically on technology, tests work through “projection”:

If a scale model of a Boeing 747 airfoil performs satisfactorily in a wind tunnel, we can project that the wing of a Boeing 747 will perform satisfactorily in actual flight … It is the assumption of this similarity relationship that enables the projection to be made and that enables engineers warrantably to use the test results as grounds that they have found out something about the actual working of the technology (1993: 29).

The connection with a predictive mechanism is clear here, as projection entails not being surprised when we move into the new context of the “actual world” having specified certain relationships in the “test world.” The projection/predictive aspect is made almost verbatim here: “In order to say two things are similar, we bracket, or place in abeyance, all the things that make for possible differences. In other words, we select from myriad possibilities the relevant properties whereby we judge two things to be similar … [The] outcome of the tests can be taken to be either a success or a failure, depending upon the sorts of similarity and difference judgments made” (32).

Thus, a generative model is made in the testing environment, and it is then applied in the actual world environment on the understanding that we will not need to identify predictive error when we do this, as the generative model is similar enough to the actual world that we will have already resolved those. As Pinch concludes, “The analysis of testing developed here is, I suggest, completely generalizable. The notion of projection and the similarity relationships that it entails are present in all situations in which we would want to talk about testing” (37). And, it does seem that this particular use of testing can find analogues far and wide, including with the laboratory testing that is Latour’s focus and more generally we might say with educational or vocational testing where, likewise, a similarity relationship depends on a test that can minimize the difference between two contexts (a difference that we can understand according to the presence, or hopefully lack thereof, of prediction error). But what if we try to apply the test concept to something more remote from science and technology, like morality?

On this front, we can find statements like the following, from Boltanski and Thevenot:

A universe reduced to a common world would be a universe of definite worths in which a test, always conclusive (and thus finally useless), could absorb the commotion and silence it. Such an Eden-like universe in which ‘nothing ever happens by chance’ is maintained by a kind of sorcery that exhausts all the contingencies … An accident becomes a deficiency … Disturbed situations are often the ones that lead to uncertainties about worth and require recourse to a test in order to be resolved. The situation is then purified … In a true test, deception is unveiled: the pea under the mattress discloses the real princess. The masks fall; each participant finds his or her place. By the ordering that it presupposes, a peak moment distributes the beings in presence, and the true worth of each is tested (2006: 136-138).

In this rendering, tests are quite explicitly meant to make “accidents” stand out, in addition to fraud and fakery. The goal is the construction of a situation removed of all contingencies, in which, likewise, we do not notice anything because the test has put it in its proper order. When we do notice certain things (e.g. “the same people win all the same tests,” “they are singled out unfairly,” “they never got the opportunity,” etc), these are prediction errors based on some predictive ordering of the world that creates expectation. Simultaneously they are meaningful (for people) as forms of injustice.

Boltanski and Thevenot dovetail, on this point, with something that became clear for at least one person in the tradition of probability theory, namely Blaise Pascal (see Daston 1988: 15ff). For Pascal, the expectations formed by playing a game of chance could themselves be the source of noticing the equivalent of “error,” for instance, when some player wins far too often while another never wins. A test is the source of an order “without contingency” where “nothing ever happens by chance,” which in this case means a test is the rules of the game that allow for possibilities (all can win) while resolving those possibilities into a result (only one will win). This creates expectations, and Boltanski and Thevenot extrapolate from this (citing sports contests as epitomizing their theory) to identify “worlds” as different versions of this predictive ordering. Injustice is officially revealed at a second level of testing, then, as the test that creates this order can itself be tested (see Potthast 2017). Prediction errors can be noticed, likewise these can be resolved through the adaptation of a generative model, which would seem to demand a reformative (or revolutionary) change of the test in a manner that would subsequently allow it to meet expectations.

A genealogy of testing

What is interesting about these examples is, abstracted from history as they are, they demonstrate parallel wings of a tradition that Foucault traces to the decline of the “ordeal” and the birth of the “inquiry.” Both of these fit the profile of the test, though only the former gives the outcome the kind of official status or legitimacy of the laboratory test, the technological test, or the moral test. The ordeal involves a sheer confrontation that can occur at any time, and which creates expectations strictly in relation to some other specific thing, whether this be another person or something inanimate and possibly dangerous (like fire) or a practice of some kind (like writing a book). One can always test themselves against this again, and to move beyond known limitations, they must test themselves if they are to do anything like revise a generative model by encountering different prediction errors.

Foucault’s larger point here recommends a more general argument, rooted in a kind of genealogy, that justice requires a caraceral; that the only form of justice is the one that rests in illegality. On the contrary, in his earlier work Foucault recommends a different approach to justice, one that renders any necessary association of justice and “the carceral archipelago” mistaken, as it would only consist of a relatively recent, though impactful, appropriation of justice. Thus, the argument Foucault presents is less nominal than it may seem at first, particularly when we consider the following:

What characterizes the act of justice is not resort to a court and to judges; it is not the intervention of magistrates (even if they had to be simple mediators or arbitrators). What characterizes the juridical act, the process or the procedure in the broad sense, is the regulated development of a dispute. And the intervention of judges, their opinion or decision, is only ever an episode in this development. What defines the juridical order is the way in which one confronts one another, the way in which one struggles. The rule and the struggle, the rule in the struggle, this is the juridical (Foucault 2019: 116).

Here the meaning of justice is expanded to refer to the “regulated development of a dispute,” which may or may not have judges, which may or may not take place in a court, result in a judgment, or find at its culmination some sort of definitive decision or “judgment.” All of these are added features to the basic dispute.

Elsewhere Foucault expands on this by changing the language he uses in a significant way: from “dispute” justice shifts to “trial,” which he gives this an expansive meaning by drawing a distinction within the category of trial itself and distinguishing between epreuve and inquiry. There is a historical tension in the distinction: inquiries will come to replace epreuves (or “ordeals”) in a Eurocentric history. This division is apparent as early as the ancient Greeks who, in a Homeric version, would create justice through the rule-governed dispute, with the responsibility for deciding—not who spoke the truth, but who was right–entrusted to the fight, the challenge, and “the risk that each one would run.” Contrary to this the Oedipus Rex form, as exemplified by Sophocles’ great play. Here, in order to resolve a dispute of apparent patricide, we find one of the emblems of Athenian democracy: “the people took possession of the right to judge, of the right to tell the truth, to set the truth against their own masters, to judge those who governed them” (Foucault 2000: 32-33).

This division would be replicated in the later distinctions of Roman law, as rooted inquiry, and Germanic law, as rooted in something more resembling the contest or epreuve, with disputes conducted through either means. Yet with the collapse of the Carolingian Empire in the tenth century, “Germanic law triumphed, and Roman law fell into oblivion for several centuries.” Thus, feudal justice consisted of “disputes settled by the system of the test,” whether this be a “test of the individual’s social standing,” a test of verbal demonstration in formulaically presenting the grievance or denunciating one another, tests of an oath in which “the accused would be asked to take an oath and if he declined or hesitated he would lose the case,” and finally “the famous corporal, physical tests called ordeals, which consisted of subjecting a person to a sort of game, a struggle with his own body, to find out whether he would pass or fail.”

As the trajectory of justice moves, then, the role and place of the epreuve ascends to prominence; testing becomes justice, in other words, as the means to resolve a dispute centers around the ordeal and its outcome, more generally as a way of letting God’s voice speak. In one general account, the trial by “cold water” involved “dunking the accused in a pond or a cistern; if the person sank, he or she was pronounced innocent, and if the person floated, he or she was found guilty and either maimed or killed.” In the trial by “hot iron,” the accused would “carry a hot iron a number of paces, after which the resulting wound was bandaged. If the wound showed signs of healing after three days, the accused was declared innocent, but if the wound appeared to be infected, a guilty verdict ensued” (Kerr, Forsyth and Plyey 1992).

The epreuve, in this case, remains a trial of force or between forces, which may be codified and regulated as the case may be, as water or iron would be blessed before the ordeal, and therefore made to speak the word of God. More generally, to decline the test was to admit guilt in this binary structure, and this carried into the challenge by another in a dispute to a contest. Thus, justice ended in a victory or a defeat, which appeared definitive, and this worked in an almost “automatic” way, because it required no third party in the form of one who judges.

Across this genealogy, we find something equivalent to the creation of uncertainty, in some cases deliberately made, in other cases not, and then its resolution by some means into an outcome after a given duration of time. This outcome may have an institutional sanction (as “justice”) or it could have something more like the sanction of a fight, and presumably the certainty of what would happen should a fight happen again. In these different ways, predictions are made and expectations settled. An “error” stands out as noticeable in a variety of forms: as someone with whom one has a dispute, as an action taken or event that happened but was not expected, whether according to explicitly defined rules or not, or in the case of the democratic link suggested by Foucault, the pressing question of who should rule and whether such rule can be legitimate (see Mouffe 2000).

Some equivalent to the test (whether as inquiry or ordeal) is involved in all of these cases, and in the genealogy at least, we can glimpse how consequential it might be for a new test form to come on the scene, or to win out over another, as a way of, in a sense, appropriating cognitive susceptibilities that must be activated should “testing” make any difference for predictive dispositions.

Conclusion

The larger point is that the concept of test is substantive, here, because we can bridge its properties to properties of cognition. The task is to say that the predictive dispositions that are cognitive create a susceptibility to tests: more specifically, we are likely to find tests meaningful because of our predictive dispositions. If tests are drawn upon across all of these different areas, specifically in cases of uncertainty (whether as dispute, as experiment, as how to design a technology) or what we have established in general terms as “situations in which we are presently engaged with prediction error that we cannot help but notice a lot,” then it would follow that we are susceptible to tests as what allows us to absorb this uncertainty, a process we cannot understand or even fully recognize without reference to “real features of real brains” (Turner 2007). This, I want to propose, is how we can approach “test” as a dual reference concept, and its applicability in areas as varied as religion, politics, science, morality, and technology.

Tests are “meaningful for people” when they absorb uncertainty and generate expectation. They are also meaningful for people when they create uncertainty and enable critique. We could not identify something like a “test” if tests did not have these kinds of cognitive effects, and we cannot understand those cognitive effects without finding a distinguishably cognitive process (e.g. “psychologically real” with lots of “mentalistic content” extending even to neurons). In this case, the parallel of testing and uncertainty and predictive processing and prediction error is not a distant analogy, as is often the case with bracketing concepts. To understand testing’s absorption of uncertainty we need predictive processing, but to understand how predictive processing might matter for the things sociologists care about we need testing.

I’ll conclude with the suggestion that if “test” can qualify as this sort of dual reference concept then we should favor it over other potential concepts that can account for meaning (e.g. “categories,” “worldview,” “interpretation”) but, arguably, cannot be dual reference.

Something that looks like endnotes

[1] The French “pragmatists” are, in centering “test” in their concept-formation, not to be received as illegitimate appropriators of that title. Peirce (1992) himself encouraged a focus on the study of “potential” as referring to something “indeterminate yet capable of determination in any special case.” This could very well serve as clarified restatement of the definition of test. Dewey (1998) makes the connection more explicit in his thorough conceptualization of test: “The conjunction of problematic and determinate characters in nature renders every existence, as well as every idea and human act, an experiment in fact, even though not in design. To be intelligently experimental is but to be conscious of this intersection of natural conditions so as to profit by it instead of being at its mercy. The Christian idea of this world and this life as a probation is a kind of distorted recognition of the situation; distorted because it applied wholesale to one stretch of existence in contrast with another, regarded as original and final. But in truth anything which can exist at any place and at any time occurs subject to tests imposed upon it by surroundings, which are only in part compatible and reinforcing. These surroundings test its strength and measure its endurance … That stablest thing we can speak of is not free from conditions set to it by other things … A thing may endure secula seculorum and yet not be everlasting; it will crumble before the gnawing truth of time, as it exceeds a certain measure. Every existence is an event.”

References

Bloch, Maurice (2012). Anthropology and the Cognitive Challenge. Cambridge UP.

Boltanski, Luc and Laurent Thevenot. (2006). On Justification. Princeton UP.

Bourdieu, Pierre. (1991). The Political Ontology of Martin Heidegger. Stanford UP.

Daston, Lorraine. (1988). Classical Probability in the Enlightenment. Princeton UP.

Daly, Robert. (1977). “The Soteriological Significance of the Sacrifice of Isaac.” The Catholic Biblical Quarterly 39: 45-71.

Deleuze, Gilles and Guattari, Felix. (1995). What is Philosophy? Columbia UP.

Foucault, Michel. (2019). Penal Theories and Institutions: Lectures at the College de France, 1971-72, edited by Bernard Harcourt. Palgrave.

Foucault, Michel. (2000). “Truth and Juridical Forms” in Power: The Essential Works of Michel Foucault, 1954-1984, edited by James D. Faubion. The New Press.

Geertz, Clifford. (1973). “The Growth of Culture and the Evolution of Mind” in Interpretation of Cultures.

Hutto, Daniel. (2018). “Getting into predictive processing’s great guessing game: Bootstrap heaven or hell?” Synthese 195: 2445-2458.

Kerr, Margaret, Forsyth, Richard, and Michel Plyey. (1992). “Cold Water and Hot Iron: Trial by Ordeal in England.” Journal of Interdisciplinary History 22: 573-595.

Kurkian, Dmitry. (2020). “Culture and Cognition: the Durkheimian Principle of Sui Generis Synthesis vs. Cognitive-Based Models of Culture.” American Journal of Cultural Sociology 8: 63-89.

Latour, Bruno. (1988). The Pasteurization of France. Harvard UP.

Latour, Bruno. (1999). Pandora’s Hope. Harvard UP.

Lemieux, Cyril. (2008) “Scene change in French sociology?” L’oeil Sociologique

Lizardo, Omar. (2014). “Beyond the Comtean Schema: The Sociology of Culture and Cognition Versus Cognitive Social Science.” Sociological Forum 29: 983-989.

Marres, Noortje and David Stark. (2020). “Put to the Test: For a New Sociology of Testing.” British Journal of Sociology 71: 423-443.

Mast, Jason. (2020). “Representationalism and Cognitive Culturalism: Riders on Elephants on Turtles All the Way Down.” American Journal of Cultural Sociology 8: 90-123.

Marcel, Mauss. (1966). The Gift. Something UP.

Menary, Richard. (2015). “Pragmatism and the Pragmatic Turn in Cognitive Science” in The Pragmatic Turn: Toward Action-Oriented Views in Cognitive Science. MIT Press.

Mouffe, Chantal. (2008). The Democratic Paradox. Verso.

Norton, Matthew. (2018). “Meaning on the Move: Synthesizing Cognitive and Systems Concepts of Culture.” American Journal of Cultural Sociology 7: 1-28.

Pinch, Trevor. (1993). “Testing—One, Two, Three… Testing!”: Toward a sociology of testing. Science, Technology, & Human Values, 18(1), 25–41.

Potthast Jorg. (2017) The sociology of conventions and testing. In: Benzecry C, Krause M and Reed IA (eds) Social Theory Now. Chicago: University of Chicago Press, 337–361.

Popper, Karl. (1997). The Logic of Scientific Discovery. Routledge.

Ronnell, Avital. (2005). The Test Drive. University of Illinois Press.

Sewell, William. (2005). “History, Synchrony, and Culture: Reflections on the Work of Clifford Geertz” in Logics of History.

Sperber, Dan. (1985). “Anthropology and Psychology: Towards an Epidemiology of Representations.” Man 20: 73-89.

Strand, Michael. (2020). “Sociology and Philosophy in the United States since the Sixties: Death and Resurrection of a Folk Action Obstacle.” Theory and Society 49: 101-150.

Strand, Michael (2021). “Cognition, Practice and Learning in the Discourse of the Human Sciences” in Handbook in Classical Sociological Theory. Springer.

Strand, Michael and Omar Lizardo. (forthcoming). “For a Probabilistic Sociology: A History of Concept-Formation with Pierre Bourdieu” Theory and Society

Strand, Michael and Omar Lizardo (forthcoming). “Chance, Orientation and Interpretation: Max Weber’s Neglected Probabilism and the Future of Social Theory.” Sociological Theory

Turner, Stephen. (2007). “Social Theory as Cognitive Neuroscience.” European Journal of Social Theory 10: 357-374.

From Dual-Process Theories to Cognitive-Process Taxonomies

Although having a history as old as the social and behavioral sciences (and for some, as old as philosophical reflections on the mind itself), dual-process models of cognition have been with us only for a bit over two decades, becoming established in cognitive and social psychology in the late 1990s (see Sloman, 1996 and Smith and DeCoster, 2000 for foundational reviews). The implicit measurement revolution provided the “data” side to the theoretical and computational modeling side, thus fomenting further theoretical and conceptual development (Strack & Deutsch, 2004; Gawronski & Bodenhausen, 2006). Although not without its critics, the dual-process approach has now blossomed into an interdisciplinary framework useful for studying learning, perception, thinking, and action (Lizardo et al., 2016). In sociology, dual-process ideas were introduced by way of the specific dual-process model of moral reasoning developed by Jonathan Haidt (2001) in Steve Vaisey’s (2009) now classic and still heavily cited paper. Sociological applications of the dual-process framework for specific research problems now abound, with developments on both the substantive and measurement sides (Miles, 2015; Miles et al. 2019; Melamed et al. 2019; Srivastava & Banaji, 2011).

The dual-process framework revolves around the ideal-typical distinction between two “modes” or “styles” of cognition (Brett & Miles, 2021). These are now very familiar. One is the effortful, usually conscious, deliberate processing of serially presented information, potentially available for verbal report (as when reasoning through a deductive chain or doing a hard math problem in your head). The other is the seemingly effortless, automatic, usually unconscious, associative processing information (as when a solution to a problem just “comes” to you, or when you just “know” something without seemingly having gone through steps to reach the solution). This last is usually referred to as intuitive, automatic, or associative “Type 1” cognition, and the former is usually referred to as effortful, deliberate, or non-automatic “Type 2” cognition.

As with many hard and fast distinctions, there is the virtue of simplification and analytic power, but there is the limitation, evident to all, that the differentiation between Type 1 versus Type 2 cognition occludes as much as it reveals. For instance, people wonder about the existence of “mixed” types of cognition or iterative cycles between the two modes or the capacity of one mode (usually Type 2) to override the outputs of the other (usually Type 1). It seems like the answer to all these wonders is a general “yes.” We can define a construct like “automaticity” to admit various “in-between” types (Moors & De Houwer, 2006), suggesting that a pure dichotomy is too simple (Melnikoff & Bargh, 2018). And yes, the two types of cognition interact and cycle (Cunningham & Zelazo, 2007). The interactive perspective is even built into some measurement strategies, which rely on overloading or temporarily overwhelming the deliberate system to force people to respond with intuitive Type 1 cognition (as in so-called “cognitive load” techniques; see Miles, 2015 for a sociological application).

Another sort of wonder revolves around whether these are the only types of cognition that exist. Are there any more types? Accordingly, some analysts speak of “tri” or “quad” process models and the like (Stanovich, 2009). It seems, therefore, that field is moving toward a taxonomic approach to the study of cognitive processes. However, the criteria or “dimensions” around which such taxonomies are to be constructed are in a state of flux. As I noted in a previous post, moving toward a taxonomic approach is generally a good thing. Moreover, the field of memory research is a good model for how to build taxonomic theory in cognitive social science (CSS), especially since the “kinds” typically studied in CSS are usually “motley” (natural kinds that decompose into fuzzy subkinds). When studying motley kinds and organizing into fruitful taxonomies, it is essential to focus on the analytic dimensions and let the chips fall where they may. This is different from thinking up “new types” of cognition from the armchair in unprincipled ways, where the dimensions that define the types are ill-defined (as with previous attempts to talk about tri-process models of cognition and the like). Moreover, the dimensional approach leaves things open to discover surprising “subkinds” that join properties that we would consider counter-intuitive.

Accordingly, an upshot of everyone now accepting (even begrudgingly) some version of the dual-process theory is that we also agree that the cognitive-scientific kind “cognition” is itself motley! That is, whatever it is, cognition is not a single kind of thing. Right now, we kind of agree that it is at least two things (as I said, an insight that is as old as the Freudian distinction between primary and secondary process), but it is likely that it could be more than two. In this post, I’d like to propose one attempt to define the possible dimensional space from which a more differentiated typology of cognitive processes can be constructed.

Taxonomizing Cognition

So if we needed to choose dimensions to taxonomize cognition, where would we begin? I think a suitable candidate is to pick two closely aligned dimensions of cognition that people thought were fused or highly correlated but now are seen as partially orthogonal. For example, in a previous post on the varieties of “implictness” (which is arguably the core dimension of cognition that defines the core distinction in dual-process models), I noted that social and cognitive psychologists differentiate between two criteria for deeming something “implicit.” First, a-implicitness uses an “automaticity” criterion. Here, cognition is implicit if it is automatic and explicit if it is deliberate or effortful. Second, there is u-implicitness, which uses a(n) (un)consciousness criterion. Here, cognition is implicit if it occurs outside of consciousness, and it is explicit if it is conscious.

I implied (but did not explicitly argue) in that post that maybe these two dimensions of explicitness could come apart. If they can, these seem like pretty good criteria to build a taxonomy of cognitive process kinds that goes beyond two! This is precisely what the philosophers Nicholas Shea and Chris Frith did in a paper published in 2016 in Neuroscience of Consciousness. Cross-classifying the type of processing (deliberate v. automatic) against the type of representations over which the processing occurs (conscious v. unconscious), yields a new “type” of cognition which they refer to as “Type 0 cognition.”

In Shea and Frith’s taxonomy, our old friend Type 1 cognition refers to the automatic processing of initially conscious representations, typically resulting in conscious outputs. In their words, “[t]ype 1 cognition is characterized by automatic, load-insensitive processing of consciously represented inputs; outputs are typically also conscious.” (p.4). This definition is consistent with Evans’s (2019) more recent specification of Type 1 cognition as working-memory independent cognition that still uses working memory to “deposit” the output of associative processing. In Evans’s words,

While Type 1 processes do not require the resources of working memory or controlled attention for their operation (or they would be Type 2) they do post their products into working memory in a way that many autonomous processes of the brain do not. Specifically, they bring to mind judgements or candidate responses of some kind accompanied by a feeling of confidence or rightness in that judgement (p. 384).

For Shea and Frith (2016), on the other hand, our other good friend, Type 2 cognition, refers to the deliberate, effortful processing of conscious representations. In their words,

Type 2 cognition is characterized by deliberate, non-automatic processing of conscious representations. It is sensitive to cognitive load: type 2 processes interfere with one another. Type 2 cognition operates on conscious representations, typically in series, over a longer timescale than type 1 cognition. It can overcome some of the computational limitations of type 1 cognition, piecemeal, while retaining the advantage of being able to integrate information from previously unconnected domains. It is computation-heavy and learning-light: with its extended processing time, type 2 cognition can compute the correct answer or generate optimal actions without the benefit of extensive prior experience in a domain (p. 5).

By way of contrast with these familiar faces, our new friend Type 0 cognition refers to the automatic processing of non-conscious representations. Shea and Frith see isolating Type 0 cognition as a separate cognitive-process subkind as their primary contribution. Previous work, in their view, has run Type 0 and Type 1 cognition together, to their analytic detriment. Notably, they argue for the greater (domain-specific) efficiency and accuracy of Type 0 cognition over Type 1. They note that various deficiencies of Type 1 cognition identified in such research programs as the “heuristics and biases” literature come from the fact that, in Type 1 cognition, there is a mismatch between process and representation because automatic/associative processes are recruited to deal with conscious representational inputs.

For instance, Type 1 cognition is at work when Haidt asks people whether they would wear Hitler’s t-shirt, and they say “ew, no way!” but are unable to come up with a morally reasonable reason why (or make up an implausible one on the spot). Type 1 moral cognition “misfires” here because the associative (“moral intuition”) system was recruited to process conscious inputs, relied on an associative/heuristic process to generate an answer (in this case, based on implicit contact, purity, and contagion considerations), and produced a conscious output, the origins of with subjects are completely unaware of (and is thus forced to retrospectively confabulate using Type 2 cognition). The same goes for judgment and decision-making producing answers to questions when engaging in the base-rate fallacy, using a representativeness heuristic, and the like (Kahneman, 2011).

The types of cognition for which a match is made in heaven between process and representation (like Type 2 and their Type 0) result in adaptive cognitive processes that “get the right answer.” Type 2 cognition refers to domain-general problems requiring information integration and the careful weighing of alternatives. In Type 0 cognition, this refers to domain-specific problems requiring fast, adaptive cognitive processing and action control, where consciousness (if it were to rear its ugly head) would spoil the fun and impair the effectiveness of the cognitive system to do what is supposed to do, similar to athletes who “choke” when they become conscious of what they are doing (see Beilock, 2011).

So, what is Type 0 cognition good for? Shea and Frith point to things like the implicit learning of probabilistic action/reward contingencies after many exposures (e.g., reinforcement learning), where neither the probabilities nor the learning process is consciously represented, and the learning happens via associative steps. As they note, in “model-free reinforcement learning can generate optimal decisions when making choices for rewards, and feedback control can compute optimal action trajectories…non-conscious representation goes hand-in-hand with correct performance” (p. 3). In the same way, “Type 0 cognition is likely to play a large role in several other domains, for example in the rich inferences which occur automatically and without consciousness in the course of perception, language comprehension and language production” (ibid).

Organizing the Types

So, where does Shea and Frith’s taxonomy of cognitive process kinds leave us? Well, maybe something like the dimensional typology shown in Figure 1. It seems like at least three different cognitive process kinds are well-defined, especially if you are convinced that we should distinguish Type 0 from Type 1 cognition (and I think I am).

However, as I argued earlier, a key advantage of beginning with dimensions in any taxonomical exercise is that we may end up with a surprise. Here, it is the fact that a fourth potential type of cognition now appears in the lower-right quadrant, one that no one has given much thought to before. Type ??? cognition: deliberate processing of unconscious representations. Can this even be a thing? Shea and Frith do note this implication of their taxonomic exercise but think it is too weird. They even point out that it may be a positive contribution of their approach to have discovered this “empty” slot in cognitive-process-kind space. In their words, “[w]hat of the fourth box? This would be the home of deliberate processes acting on non-conscious representations. It seems to us that there may well be no type of cognition that fits in this box. If so, that is an important discovery about the nature of consciousness” (p. 7).

Nevertheless, are things so simple? Maybe not. The Brains Blog dedicated a symposium to the paper in 2017 in which three authors provided commentaries. Not surprisingly, some of the commenters did not buy the “empty slot” argument. In their comment, Jacob Berger points to some plausible candidates for Shea and Frith’s Type ??? cognition (referred to as “Type 0.5 cognition”). This includes the (somewhat controversial) work of Dijksterhuis, Aarts, and collaborators (e.g., Dijksterhuis & Nordgren, 2006; Dijksterhuis & Aarts, 2010) on “unconscious thought theory” (UTT) (see Bargh, 2011 for a friendly review). In the UTT paradigm, participants are asked to make seemingly deliberate choices between alternatives, with a “right” answer aimed at maximizing a set of quality dimensions. At the same time, conscious thinking is impaired via cognitive load. The key result is that participants who engage in this “unconscious thinking” end up making choices that are as optimal as people who think about it reflectively. So, this seems to be a case of a deliberate thinking process operating over unconscious representations.

Berger does anticipate an objection to UT as being a candidate for Type ??? cognition, which itself brings up an issue with critical taxonomic ramifications:

S&F might reply that such [UT] cases are not genuinely unconscious because, like examples of type-1 cognition, they involve conscious inputs and outputs. But if this processing is not type 0.5, then it is hard to see where S&F’s taxonomy accommodates it. The cognition does not seem automatic, akin to the processing of type 0 or type 1 of which one is unaware (it seems, for example, rather domain general); nor does it seem to be a case of type-2 cognition, since one is totally unaware of the processing that results in conscious outputs. Perhaps what is needed is an additional distinction between the inputs/outputs of a process’ being conscious and the consciousness of states in the intervening processing. In type-1 cognition, the inputs/outputs are conscious, but the states involved in the automatic processing are not; in type-2, both are conscious. We might therefore regard Dijksterhuis’ work as an instance of ‘type-1.5’ cognition: conscious inputs/outputs, but deliberative unconscious processing.

Thus, Berger proposes to dissociate not only conscious/unconscious representations from deliberate/automatic processing but also adds the dimension of whether the inputs and outputs of the cognitive process and its intervening steps are themselves conscious or unconscious. Berger’s implied taxonomy can thus be represented as in Figure 2.

Figure 2 clarifies that the actual mystery type does not connect conscious inputs and outputs with deliberate unconscious processing (UT), but a type linking unconscious inputs and outputs with deliberate unconscious processing (the new Type ???). Also, the figure makes clear that the proper empty slot is a type of cognition conjoining unconscious inputs and outputs with deliberate conscious processing; this bizarre and implausible combination can indeed be ruled out on a priori grounds. Note, in contrast, that if there is such a thing as deliberate unconscious processing (and the jury is still out on that), there is no reason to rule out the new Type ??? cognition shown in Figure 2 on a priori grounds (as Shea and Frith tried to do with Berger’s Type 1.5). For instance, Bargh (2011) argues that unconscious goal pursuit (a type of unconscious thought) can be triggered outside of awareness (unconscious input) and also has behavioral consequences (e.g., trying hard on a task) that subjects may also be unaware of (unconscious output). In this sense, Bargh’s unconscious goal pursuit would qualify as a candidate for Type ??? cognition. So, following Berger’s recommendation, we end up with five (I know an ugly prime) candidate cognition types.

So, What?

Is all we are getting after all of this a more elaborate typology? Well, yes. And that is good! However, I think the more differentiated approach to carving the cognitive-process world also leads to some substantive insight. I refer in particular to Shea and Frith’s introduction of the Type 0/Type 1 distinction. For instance, in a recent review (and critique) of dual-process models of social cognition, Amodio proposes an “interactive memory systems” account of attitudes and impression formation (“Social Cognition 2.0”) that attempts to go beyond the limitations of the traditional dual-process model (“Social Cognition 1.0”).

Amodio’s argument is wide-ranging, but his primary point is that there are multiple memory systems and that a conception of Type 1 cognition as a single network of implicit concept/concept associations over which unconscious cognition operates is incomplete. In addition to concept/concept associations, Amodio points to other types of associative learning, including Pavlovian (affective) and instrumental (reinforcement learning). Amodio’s primary point is that something like an “implicit attitude,” insofar as it recruits multiple but distinct (and dissociable) forms of memory and learning subserved by different neural substrates, is not a single kind of thing (a taxonomical exercise for the future!). This dovetails nicely with the current effort to taxonomize cognitive processes. Thus, a standard conceptual association between categories of people and valenced traits operates via Type 1 cognition. However, it is likely that behavioral approach/avoid tendencies toward the same type of people, being the product of instrumental/reinforcement learning mechanisms, operate via Shea and Frith’s Type 0 cognition.

References

Bargh, J. A. (2011). Unconscious Thought Theory and Its Discontents: A Critique of the Critiques. Social Cognition, 29(6), 629–647.

Beilock, S. L. (2011). Choke. The secret of performing under pressure. London: Constable.

Brett, G., & Miles, A. (2021). Who Thinks How? Social Patterns in Reliance on Automatic and Deliberate Cognition. Sociological Science, 8, 96–118.

Cunningham, W. A., & Zelazo, P. D. (2007). Attitudes and evaluations: a social cognitive neuroscience perspective. Trends in Cognitive Sciences, 11(3), 97–104.

Dijksterhuis, A., & Aarts, H. (2010). Goals, attention, and (un)consciousness. Annual Review of Psychology, 61, 467–490.

Evans, J. S. B. T. (2019). Reflections on reflection: the nature and function of type 2 processes in dual-process theories of reasoning. Thinking & Reasoning, 25(4), 383–415.

Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: an integrative review of implicit and explicit attitude change. Psychological Bulletin, 132(5), 692–731.

Haidt, J. (2001). The emotional dog and its rational tail: a social intuitionist approach to moral judgment. Psychological Review, 108(4), 814–834.

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.

Lizardo, O., Mowry, R., Sepulvado, B., Stoltz, D. S., Taylor, M. A., Van Ness, J., & Wood, M. (2016). What are dual process models? Implications for cultural analysis in sociology. Sociological Theory, 34(4), 287–310.

Melamed, D., Munn, C. W., Barry, L., Montgomery, B., & Okuwobi, O. F. (2019). Status Characteristics, Implicit Bias, and the Production of Racial Inequality. American Sociological Review, 84(6), 1013–1036.

Melnikoff, D. E., & Bargh, J. A. (2018). The mythical number two. Trends in cognitive sciences, 22(4), 280-293.

Miles, A. (2015). The (Re)genesis of Values: Examining the Importance of Values for Action. American Sociological Review, 80(4), 680–704.

Miles, A., Charron-Chénier, R., & Schleifer, C. (2019). Measuring Automatic Cognition: Advancing Dual-Process Research in Sociology. American Sociological Review, 84(2), 308–333.

Moors, A., & De Houwer, J. (2006). Automaticity: a theoretical and conceptual analysis. Psychological Bulletin, 132(2), 297–326.

Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119(1), 3.-22

Smith, E. R., & DeCoster, J. (2000). Dual-Process Models in Social and Cognitive Psychology: Conceptual Integration and Links to Underlying Memory Systems. Personality and Social Psychology Review: An Official Journal of the Society for Personality and Social Psychology, Inc, 4(2), 108–131.

Srivastava, S. B., & Banaji, M. R. (2011). Culture, Cognition, and Collaborative Networks in Organizations. American Sociological Review, 76(2), 207–233.

Stanovich, K. E. (2009). Distinguishing the reflective, algorithmic, and autonomous minds: Is it time for a tri-process theory? In J. S. B. T. Evans (Ed.), In two minds: Dual processes and beyond , (pp (Vol. 369, pp. 55–88). Oxford University Press.

Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review: An Official Journal of the Society for Personality and Social Psychology, Inc, 8(3), 220–247.

Vaisey, S. (2009). Motivation and Justification: A Dual-Process Model of Culture in Action. American Journal of Sociology, 114(6), 1675–1715.

Consciousness and Schema Transposition

In a recent paper published in American Sociological Review, Andrei Boutyline and Laura Soter bring much-needed conceptual clarification to the sociological appropriation of the notion of schemas while also providing valuable and welcome guidance on future uses of the concept for practical research purposes. The paper is a tour de force, and all of you should read it (carefully, perhaps multiple times), so this post will not summarize their detailed argument. Instead, I want to focus on a subsidiary but no less important set of conclusions towards the end, mainly having to do with the relationship between declarative and nondeclarative cognition and an old idea in sociological action theory due to Bourdieu (1980/1990) that was further popularized in the highly cited article by Sewell (1992) on the duality of structure. I refer to the notion of schematic transposition.

In what follows, I will first outline Bourdieu’s and Sewell’s use of the notion and then go over how Boutyline and Soter raise a critical technical point about it, pointing to what is perhaps a consequential theoretical error. Finally, I will close by pointing to some lines of evidence in cognitive neuroscience that seem to buttress Boutyline and Soter’s position.

The idea of schematic transposition is related to an older idea due to Piaget of schema transfer. The basic proposal is that we can learn to engage in a set of concrete activities (e.g., let’s say “seriation” or putting things in rows or lines) in one particular practical context (putting multiple pebbles or marbles in a line). Then, after many repetitions, we develop a schema for it. Later, when learning about things in another context, let’s say “the number line” in basic arithmetic, we understand (assimilate) operations in this domain in terms of the previous seriation schema. Presumably, analogies and conceptual metaphors also depend on this schema transfer mechanism. In Logic of Practice, Bourdieu built this dynamic capacity for schema transfer into the definition of habitus everyone loves to hate, noting that the habitus can be thought of as “[s]ystems of durable, transposable dispositions, structured structures predisposed to function as structuring structures…” and so forth (p. 53).

This idea of transposibility ends up being essential for a habit theory like Bourdieu’s because it adds much-needed flexibility and creativity to how we conceive the social agent going about their lives (Joas, 1996). This is because thinking of action as driven by habitus does not entail people stuck with “one-track” inflexible or mechanical dispositions. Instead, via their capacity to transpose classificatory or practical habits learned in one domain to others, their internalized practical culture functions in a more “multi-track” way, being thus adaptive and creative. In an old paper on the notion of habitus (2004), I noted something similar to this, pointing out that “it is precisely this idea of flexible operations that allows for the habitus to not be tied to any particular content…instead, the habitus is an abstract, non-context specific, transposable matrix” (p. 391-392). Thus, there is something about transposability that seems necessary in a theory of action so that it does not come off as overly deterministic or mechanical.

In his famous 1992 paper, Sewell went even further, putting transposability at the very center of his conception of social change and agency. Departing from a critique of Bourdieu, Sewell noted two things. First (p. 16), any society contains a multiplicity of “structures” (today, we’d probably use the term “field,” “sphere,” or “domain”). Secondly (p. 17), this means people need to navigate across them somehow. Single-track theories of habit and cognition cannot explain how this navigation is possible. This navigation is made possible, according to Sewell, only by theorizing “the transposability of schemas.” As Sewell notes:

…[T]he schemas to which actors have access can be applied across a wide range of circumstances…Schemas were defined above as generalizable or transposable procedures applied in the enactment of social life. The term “generalizable” is taken from Giddens; the term “transposable,” which I prefer, is taken from Bourdieu…To say that schemas are transposable, in other words, is to say that they can be applied to a wide and not fully predictable range of cases outside the context in which they are initially learned…Knowledge of a rule or a schema by definition means the ability to transpose or extend it-that is, to apply it creatively. If this is so, then agency, which I would define as entailing the capacity to transpose and extend schemas to new contexts, is inherent in the knowledge of cultural schemas that characterizes all minimally competent members of society (p. 17-18).

Thus, in Sewell, the very concept of agency becomes defined by the actor’s capacity to transpose schemas across contexts and domains!

Nevertheless, is the link between the idea of schema and that of schematic transposition cogent? Boutyline and Soter (2021) incisively point out that it may not be. To see this, it is important to reiterate their “functional” definition of schemas as “socially shared representations deployable in automatic cognition” (735). The key here is “automatic cognition.” As I noted in an earlier post on “implicit culture,” a common theoretical error in cultural theory consists of taking the properties of forms of “explicit” representations we are familiar with and then postulating that there are “implicit” forms of representation having the same properties, except that they happen to be unconscious, tacit, implicit and the like. The problem is that representations operating at the tacit level need not (and usually cannot) share the same properties as those operating at the explicit level.

Boutyline and Soter note a similar tension in ascribing the property “transposable,” to a tacit or nondeclarative form of culture like a schema, which generally operates in type I cognition. In their words,

A..correlate of Type I cognition is domain-specificity. Type II knowledge can be context-independent and abstract—qualities enabled in part via the powerful expressive characteristics of language—and tied to general-purpose intelligence and logical or hypothetical reasoning…In contrast, Type I knowledge is often domain-specific—thoroughly tied to, and specifically functioning within, contexts closely resembling the one in which it was learned…Type II knowledge (e.g., mathematical or rhetorical tools) can be transposed with relative ease across diverse contexts, but the principles that underlie Type I inferences may not be transferrable to other domains without the help of Type II processes.

So, it seems like both Bourdieu and Sewell (drawing on Bourdieu) made a crucial property conjunction error, bestowing a magical power (transposability) to implicit (personal) culture. This type of personal culture cannot display the transposability property precisely because it is implicit (previously, I argued that people do this with a version of symbolic representational status). Boutyline and Soter (p. 742) revisit Sewell’s example of the “commodity schema,” convincingly demonstrating that, to the extent that this schema ends up being “deep” because it is transposable, specific episodes of transposability cannot themselves operate in automatic autopilot. Instead, “novel instance[s] of commodification” must be “consciously and intentionally devised” (ibid). Thus, to the extent that they are automatically deployable, schemas are non-transposable. Transposability of schemas requires that they be “representationally redescribed” (in terms of Karmiloff-Smith 1995) into more flexible explicit formats. Tying this insight to recent work on the sociological dual-process model, Boutyline and Soter conclude that the “application of existing knowledge to new domains understood as a feature of effortful, controlled cognition” (750).

Boutyline and Souter’s compelling argument does pose a dilemma and a puzzle. The dilemma is that a really attractive theoretical property of schemas (for Bourdieu, Sewell, and the many, many people who have used their insights and been influenced by their formulation) was transposability. Without it, it seems like schemas become a much diminished and less helpful concept. The puzzle is that there are many historical and contemporary examples of empirical instances of what looks like schematic transposition. How does this happen?

Here, Boutyline and Soter provide a very elegant theoretical solution, drawing on recent work suggesting that culture can “travel” within persons across the declarative/nondeclarative divide via redescription processes and across the public/personal one via internalization/externalization processes. They note that because schemas are representational, they can be externalized (or representationally redescribed) into explicit formats (from nondeclarative to declarative). People can also internalize them from the public domain when they interact in the world (from public to personal/nondeclarative; see Arseniev-Koehler and Foster, 2020). As Boutyline and Soter note, representational redescription,

…could make the representational contents of a cultural schema available to effortful conscious cognition, which we suspect may be generally necessary to translate these representations to novel domains. After they are transformed to encompass new settings, the representational contents could then travel the reverse pathway, becoming routinized through repeated application into automatic cognition. The end product of this process would be a cultural schema that largely resembles the original schema but now applies to a broader set of domains. Representational redescription may thus be key to social reproduction, wherein familiar social arrangements backed by widely shared cultural schemas…are adapted so they may continue under new circumstances (751).

Does cognitive neuroscience’s current state of the art support the idea that consciousness is required to integrate elements from multiple experiential and cultural domains? The answer seems to be a qualified “yes,” with the strongest proponents suggesting that the very function of consciousness and explicit processing is cross-domain information integration (Tononi, 2008). A more plausible weaker hypothesis is that consciousness greatly facilitates such integration. Without it, the task would be challenging, and for complex settings such as the socio-cultural domains of interest to sociologists, perhaps impossible. As noted by the philosophers Nicholas Shea and Chris Frith,

The role of consciousness in facilitating information integration can be seen in several paradigms in which local regularities are registered unconsciously, but global regularities are only detected when stimuli are consciously represented…consciousness makes representations available to a wider range of processing, and processing that occurs over conscious representations takes a potentially wider range of representations as input (2016, p. 4).

This account supports Boutyline and Soter’s insightful observation that it was an initial mistake to link the property of transposability to schemas, especially in the initial formulation by Bourdieu, where schemas were seen as part of habitus (Vaisey, 2009). Therefore, schemas reside in the implicit mind and operate as automatic Type I cognition (Sewell was more ambiguous in this last respect). Work in cognitive psychology and the cognitive neuroscience of consciousness supports the idea that transposition requires information integration across domains. For complex domains, conscious representation and deliberate processing may be necessary for the initial stages of transposition (Shea & Frith, 2016). Of course, as Boutyline and Souter note, once institutional entrepreneurs have engaged in the first bout of transposition mediated by explicit representations, the new schema-domain linkage can be learned by others via proceduralization and enskilment, becoming part of implicit personal culture operating as Type I cognition.

Finally, a corollary of the preceding is that we may not want to follow Sewell in completely collapsing the general concept of agency into the more restricted idea of schematic transposition, as this would have the untoward consequence of reducing agency to conscious representations and system II processing over these, precisely the thing that practice and habit theories were designed to prevent.

References

Arseniev-Koehler, A., & Foster, J. G. (2020). Machine learning as a model for cultural learning: Teaching an algorithm what it means to be fat. In arXiv [cs.CY]. arXiv. https://doi.org/10.31235/osf.io/c9yj3

Bourdieu, P. (1990). The logic of practice (R. Nice, trans.). Stanford University Press. (Original work published 1980)

Boutyline, A., & Soter, L. K. (2021). Cultural Schemas: What They Are, How to Find Them, and What to Do Once You’ve Caught One. American Sociological Review, 86(4), 728–758.

Joas, H. (1996). The Creativity of Action. University of Chicago Press.

Karmiloff-Smith, A. (1995). Beyond Modularity: A Developmental Perspective on Cognitive Science. MIT Press.

Lizardo, O. (2004). The Cognitive Origins of Bourdieu’s Habitus. Journal for the Theory of Social Behavior, 34(4), 375–401.

Sewell, W. H., Jr. (1992). A Theory of Structure: Duality, Agency, and Transformation. The American Journal of Sociology, 98(1), 1–29.

Shea, N., & Frith, C. D. (2016). Dual-process theories and consciousness: the case for ‘Type Zero’cognition. Neuroscience of Consciousness, 2016(1).

Tononi, G. (2008). Consciousness as integrated information: a provisional manifesto. The Biological Bulletin, 215(3), 216-242.

Vaisey, S. (2009). Motivation and Justification: A Dual-Process Model of Culture in Action. American Journal of Sociology, 114(6), 1675–1715.

A Taxonomy of Artifactual (Cultural) Kinds

In previous posts, I made a broad distinction between the two “families” of cultural kinds. This distinction was based on the way they fundamentally interact with people. Some cultural kinds do their work because they can be learned or internalized by people. Other cultural kinds do their work not because people internalize them but because they can be wielded or manipulated. For the most part, these last exist outside people (or at least being potentially separable from people’s bodies). We referred to the former as cultural-cognitive kinds (or cognitive kinds for short) and to the latter as artifactual cultural kinds (or artifactual kinds for short).

Most of the cultural stuff that exists outside of people (so-called “public culture”) is either an artifact, whether simple or complex (usually referred to as “material culture”), a systematic or improvised coupling between a person and an artifact (usually mediated by an internalized cultural kind such as a learned skill or ability), or a more extended socio-material ensemble (Hutchins, 1995; Malafouris, 2013), consisting of the distributed agglomeration of artifacts, people, and the knowledge (both explicit and implicit) required to use the artifacts in the setting for particular purposes, whether instrumental, expressive, or performative. Traditional cultural theory in sociology and anthropology tends to embody purpose in internalized cultural-cognitive kinds such as beliefs, goals, and values. However, an argument can be made that nothing embodies purpose (and even teleology) more directly than artifactual kinds designed to accomplish concrete ends (Malafouris, 2013).

Subsequent posts were dedicated to the process via which people internalize cultural-cognitive kinds. These reflections yielded an emergent and intuitive typology within the broad “family” of cultural cognitive kinds. Some cognitive kinds are like beliefs, encoding explicit declarations or propositions. Other cognitive kinds are more like skills or abilities and are difficult to verbalize in explicit form. A third form is in-between, more like concepts, encoding general semantic knowledge (both schematic and detail-rich) of the explicit and implicit aspects of categories. Riffing on a classic distinction in the philosophy of mind and action, we referred to the first kind as “knowlege-that,” the second kind as “knowlege-how,” and the third one as “knowledge-what.” The idea is that this provides an admittedly rough but exhaustive taxonomy of cultural-cognitive kinds as people internalize them.

Given this, it is easy to form the impression that artifactual (public) cultural kinds are an undifferentiated mass. However, recent work in cognitive science and philosophy has endeavored to provide a more differentiated taxonomic picture of the various forms artifactual kinds can take (Fasoli, 2018; Heersmink, 2021; Viola, 2021). In a forthcoming paper in a special issue of Topics in Cognitive Science dedicated to “the cognitive science of tools and techniques,” Richard Heersmink (2021) provides a useful generic typology of artifactual cultural kinds that aims for the same level of generality and exhaustiveness, concerning artifactual cultural kinds, as the knowledge-that/how/what typology concerning cultural-cognitive kinds.

Heersmink (2021) defines an artifact in the broadest sense as “material objects or structures that are made to be used to achieve an aim.” Heersmink differentiates between four broad families of artifacts: Embodied, perceptual, cognitive, and affective. To each type of artifact corresponds a specific set of skills of abilities people develop when they become good and proficient at using them, which Heersmink refers to as techniques (an approach in the same spirit as Mauss, 1973). Thus, there are embodied techniques, perceptual techniques, and so forth.

Artifact/technique is an important distinction, which separates the “cognitive” family of cultural kinds from the artifactual one. However, they tend to be run together in the literature. For instance, Hutchins (1995, p.) refers to the internalized (ability) component corresponding to the use of an external artifact as an “internal artifact.” However, this is confusing and blurs an important analytic line. As Heersmink (2013, p. 468) noted in earlier work,

it is clarifying to make a distinction between technology and technique. A technology (or artifact) is usually defined as a physical object intentionally designed, made, and used for a particular purpose, whereas a technique (or skill) is a method or procedure for doing something. Both technologies and techniques are intentionally developed and used for some purpose and are in that sense artificial, i.e., human-made. However, it is important to note, or so I claim, that they are not both artifactual. Only technologies are artifactual in that they are designed and manufactured physical objects and in this sense what Hutchins refers to as internal artifacts, such as perceptual strategies, can best be seen as cognitive techniques, rather than as internal artifacts. Moreover, given that these cognitive techniques are learned from other navigators and are thus first external to the embodied agent, it is perhaps more accurate to refer to them as internalized cognitive techniques, rather than as internal cognitive techniques.

Being “artifactual,” and thus usable (e.g., made by people but external to people and embodied in material objects but not “internalizable” by people) is diagnostic for artifacts as public cultural kinds. In the same way, being “internalizable,” is diagnostic for cognitive kinds such as skills, know-how, and abilities. This (internalizability criterion) is the distinguishing marker that separates them from artifactual kinds. Both are cultural kinds because they are the historical product of human ingenuity and invention.

Embodied artifacts are the “prototypical” of the category since they show up mainly as tools we use to get stuff accomplished. In philosophy and social theory, “Heidegger’s hammer,” and Merleau-Ponty’s “blind person’s cane” are the standard examples. Enumerating specific exemplars of the category is of course an endless task, as it includes any material object that can be used to accomplish a goal (e.g., pencils, shovels, fly swatters, brooms, skateboards, keyboards, etc.). It also includes using objects not designed for a given function to accomplish a particular goal (as when we use a hammer as a doorstop). While the “proper function” of a hammer is to drive nails through a surface, it can also be used for a myriad of improvised goals, and the same goes for pretty much every embodied artifact. Concerning the person-artifact interface, the critical phenomenological transition with regard to embodied artifacts happens when we become proficient at using them after repeatedly interacting with them (or more commonly being taught by an expert user how to use them). This results in internalization, via either socialization or enculturation, of artifact-specific skills or abilities facilitating person-and-artifact couplings. Once this coupling is established, the artifact or tool becomes transparent. It is experienced as a natural extension of the body. Following Heidegger, artifacts that have achieved this level of transparency are referred as “equipment” (Dreyfus, 1984).

Perceptual artifacts are used to correct, enhance, extend, and in some cases substitute our natural perceptual abilities. Reading glasses or hearing aids are a standard (corrective) example and telescopes or binoculars a standard (enhancing/extending) example. Merleau-Ponty’s blind man’s cane can be thought of as an embodied artifact that becomes a perceptual artifact via cross-modal substitution; tactile information comes to play the functional role for non-sighted persons that visual information plays for sighted people via the mediation of the artifact. In some cases, perceptual artifacts can be engineered so that they can make available to us aspects of the world that are naturally inaccessible to us (e.g., lightwaves in the infrared range of the spectrum). This is a type of enhancement that goes beyond amplifying the usual range of our standard perceptual techniques.

Naturally, cognitive artifacts have received a tremendous amount of attention in cognitive science and the philosophy of mind (Clark, 2008). Heersmink defines them as “…human-made, material objects or structures that functionally contribute to performing a cognitive task” (Heersmink, 2021, p. 10). Cognitive artifacts have even been used as “intuition pumps,” to show how cognition and cognitive activity can be thought of as (sometimes) occurring “outside the head,” using artifactual vehicles (e.g., a notepad or an abacus) used by people to perform cognitive tasks such as remembering and calculating (Clark & Chalmers, 1998), yielding the hypothesis of “extended cognition.” Independently of their role in this particular line of investigation, cognitive artifacts are central to the study of culture. Cognitive artifacts such as calculators, maps, multiplication tables, computers, and the like are ubiquitous in our everyday lives, facilitating a virtually open-ended range of cognitive, navigational, and calculative activities that would be either very difficult or impossible to do without them.

Affective artifacts refer to “material…objects that have the capacity to alter the affective condition of the agent” (Piredda, 2019, p. 550). Under this definition, affective artifacts are pervasive and may even precede cognitive artifacts in human evolution (Langer, 1967). They include most of the human-designed implements for the production of expressive and aesthetic symbols (e.g., music, visual arts, poetry, and the like) such as musical instruments, as well as the product of their use such as aesthetic objects and performances. Language (typically a cognitive artifact), when used in particular ways to evoke affect and emotion, becomes an affective artifact. When used to evoke feeling and emotion in a ritual or aesthetic performance, or when the voice is used for a similar purpose in singing, people’s bodies and their effectors can become the affective artifact par excellence.

As Heersmink notes, these taxonomic distinctions do not imply that many artifacts end up being hybrids, performing multiple functions at once. Thus, many perceptual artifacts (e.g., a microscope) also perform cognitive functions. Cognitive artifacts (such as a family photograph) may bring up emotionally charged autobiographical memories, thus performing affective functions. Merleau-Ponty’s blind man’s cane, as noted, is both an embodied and a perceptual artifact. Artifacts can also be linked in chains, such that one kind of artifact helps us use another one. The most coupling is embodied artifacts and cognitive artifacts; for instance, mice and keyboards help us interact with computers as cognitive artifacts. Most artifacts as used in everyday dealings consist of such hybrids or multiple chains of artifact families.

References

Clark, A. (2008). Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford University Press,.

Clark, A., & Chalmers, D. (1998). The Extended Mind. Analysis, 58(1), 7–19.

Dreyfus, H. L. (1984). Between Technē and Technology: The Ambiguous Place of Equipment in Being and Time. Tulane Studies in Philosophy, 32, 23–35.

Fasoli, M. (2018). Substitutive, Complementary and Constitutive Cognitive Artifacts: Developing an Interaction-Centered Approach. Review of Philosophy and Psychology, 9(3), 671–687.

Hutchins, E. (1995). Cognition in the Wild. MIT Press.

Heersmink, R. (2013). A Taxonomy of Cognitive Artifacts: Function, Information, and Categories. Review of Philosophy and Psychology, 4(3), 465–481.

Heersmink, R. (2021). Varieties of artifacts: Embodied, perceptual, cognitive, and affective. Retrieved May 23, 2021, from https://philpapers.org/archive/HEEVOA.pdf

Langer, S. K. K. (1967). Mind: an essay on human feeling. Johns Hopkins Press.

Mauss, M. (1973). Techniques of the body. Economy and Society, 2(1), 70–88. (Original work published 1935)

Malafouris, L. (2013). How Things Shape the Mind: A Theory of Material Engagement. MIT Press.

Piredda, G. (2020). What is an affective artifact? A further development in situated affectivity. Phenomenology and the Cognitive Sciences, 19(3), 549–567.

Viola, M. (2021). Three Varieties of Affective Artifacts: Feeling, Evaluative and Motivational Artifacts. https://doi.org/10.3389/fpsyg.2016.00266

Habit as Prediction

In a previous post, Mike Strand points to the significant rise of the “predictive turn” in the sciences of action and cognition under the banner of “predictive processing” (Clark, 2015; Wiese & Metzinger, 2017). This turn is consequential, according to Mike, because it takes prediction and turns it from something that analysts, forecasters (and increasingly automated algorithms) do from something that everyone does as the result of routine activity and everyday coping with worldly affairs. According to Mike:

To put it simply, predictive processing makes prediction the primary function of the brain. The brain evolved to allow for the optimal form of engagement with a contingent and probabilistic environment that is never in a steady state. Given that our grey matter is locked away inside a thick layer of protective bone (e.g., the skull), it has no direct way of perceiving or “understanding” what is coming at it from the outside world. What it does have are the senses, which themselves evolved to gather information about that environment. Predictive processing says, in essence, that the brain can have “knowledge” of its environment by building the equivalent of a model and using it to constantly generating predictions about what the incoming sensory information could be. This works in a continuous way, both at the level of the neuron and synapse, and at the level of the whole organism. The brain does not “represent” what it is dealing with, then, but it uses associations, co-occurrences, tendencies and rhythms to predict what it is dealing with.

In this post, I would like to continue the conversation on the central role of prediction in the explanation of action and cognition that Mike started by linking it to some previous discussions on the nature and role of habit in action and the explanation of action (see here, here, and here). The essential point that I wish to make here is that there is a close link between habit and prediction. This claim may sound counterintuitive at first. The reason is that the primary way that habit and practice have been incorporated into contemporary action theory is by making habit, in its “repetitive” or “iterative” aspect, a phase or facet of action that looks mainly backward to the past (e.g., Emirbayer & Mische, 1998). Because prediction is necessarily future-oriented, most analysts think of it as also necessarily non-habitual and thus point to other non-habit like processes, such as Schutzian “projection,” that implies a break with habitual iteration. These analysts presume that there is a natural antithesis between habit and iteration (which at best may bring the past into the present) and anticipation of forthcoming futures.

Rethinking Habit for Prediction

The idea that habit is antithetical to prediction makes sense, as far as it goes, but only because it hews closely to a conception of habit that accentuates the “iterative” or repetitive side. But there are more encompassing conceptions of the role of habit in action that emphasize an iterative side to habit and an adaptive, and even “anticipatory” side. Here I focus on one such intellectual legacy of thinking about habit, which remains mostly unknown in contemporary action theory in sociology. It was developed by a cadre of thinkers, mainly in France, beginning in the early nineteenth century and extending into the early twentieth century. This approach to the notion of habit characteristically combined elements of Aristotelian, Roman-stoic, scholastic, British-empiricist, Scottish-commonsense, French-rationalist, and German-idealist philosophy, and then-novel developments in neurophysiology such as the work of Xavier Bichat. Its two leading exponents were Pierre Maine de Biran (1970) and the largely neglected (but see Carlisle (2010) and Sinclair (2019)) work of Félix Ravaisson (2008). These thinkers exercised a broad influence in the way habit was conceptualized in the French tradition, extending its influence into the work of the philosophers Albert Lemoine, Henry Bergson, and more notably, Maurice Merlau-Ponty (Sinclair, 2018).

The Double Law of Habit

The primary contribution of these two thinkers, especially Ravaisson, was developing the double law of habit. This was the proposal that habit (conceptualized as behavioral or environmental repetition) had “contradictory” effects on the “passive” (sensory, feeling) and the active (skill, action) faculties: “sensation, continued or repeated, fades, is gradually obscured and ends by disappearing without leaving a trace. Repeated movement [on the other hand] gradually becomes more precise, more prompt, and easier” (de Biran, 1970, p. 219)

In other words, facilitation in the realm of perception leads to “habituation,” meaning that experience becomes less capable of capturing attention. We become inured to the sensory flow, or in the case of experience that generate feelings (e.g., of pleasure, disgust, and so forth), the feelings “fade” in intensity (e.g., think of the difference between a first-year medical student and an experienced surgeon in the presence of a corpse). This is an argument that was deployed by Simmel to explain the “deadening” effect of urbanism on sensory discrimination and emotional reaction, generative of what he called the “blase attitude” in his classic essay on the “Metropolis and the Life of the Spirit.”

When it comes to action, on the other hand, habituation via repetition leads to the opposite of passivity; namely, facilitation of the activity (becoming faster, more precise, more self-assured) and the creation of an automatic disposition (e.g., triggered in partial or complete independence from a feeling of “willing” the action) equipped with its own inertia and bound to continue to its consummation unless interrupted. Habituated action “becomes more of a tendency, an inclination” (Ravaisson 2008: 51). This is the double face (or “law”) of habit.

Prediction as Attenuation

Trying to puzzle out these apparently contradictory effects of habituation led to a lot of head-scratching (and creative theorizing) both on the part of de Biran and Ravaisson and subsequent epigones like Bergson, Heidegger, Merleau-Ponty, and Ricoeur. Nevertheless, it becomes clear that a solution to the “double-law” puzzles emerges when the predictive dimension of both perception and action is brought to the fore. The case of “perceptual attenuation” considered below, for instance, provides the mechanism for the “fading” of the vibrancy of experience whenever we become proficient at canceling out the error produced by those experiences via top-down predictions (Hohwy, 2013). Here the “top” are generative hierarchical models instantiated across different layers in the cortex, and the bottom is incoming sensory stimulation from the world (where the job of the model is to infer the hidden causes of such stimulation).

That is, as experience is repeated and the distributed, hierarchical generative models tune their parameters to effectively figure out what’s coming before it comes, we begin to preemptively cancel out prediction error. Cancelation of prediction error leads to subsequent perceptual attenuation, such that incoming sensory information no longer commands (or requires) attention. The result is that attention is freed to concentrate on other more pressing things (e.g., the parts of the experience that are still producing precise error and thus demand it). In this respect, sensory and feeling attenuation is the price we pay for becoming good at predicting what the world offers. Prediction is at the basis of “passive” habituation (the first face of the double law).

Prediction as Facilitation

But what about the facilitation side? Here prediction, in the form of what is known as active inference, is also at play. However, this time, instead of prediction in the service of canceling out error from exteroceptive signals, the acquisition of skill turns into our capacity to cancel out prediction error emanating from our action in the world, for instance, via proprioceptive signals that track the sensory consequences of our activity. Repeated activity leads us to form increasingly accurate generative models of our action (the dynamic motor trajectory of our bodies and their effectors) in a particular environment. This means that we can anticipate what we are going to do before we do it, leading to the loss (via the mechanism described above) of the feeling of “effort” or even “willing” at the point of action initiation (Wegner, 2002), which is a phenomenological signature of habitual activity.

This is consistent with the idea that Parsonian “effort” rather than being the sine qua non of truly “free” action partially unmoored from its “conditions” (as the Kantian legacy led Parsons to implicitly assume) actually points to poorly performed (because badly predicted) action, in other words, action that is driven by generative models that are not very good at anticipating our next move. This is action that is at war with the environment not because it is “independent” from it, but because (due to lack of habituation an attunement to its objective structure of probabilities) is partially at war with it, and thus disconnected from its offerings (Silver, 2011).

The connection between habit and prediction becomes clear. On the one hand, repetition results in the attenuation of sensory input. While this was usually referred to as the “passive” side of the double-law, we can now see, drawing on recent work on predictive processing, that this is only a seeming passivity. At the subpersonal level, attenuation happens via the successful operation of well-honed generative models of the environmental causes of the input, working continuously to cancel out those incoming signals that they successfully predict. These models are one set of “habitual tracks” laid out by our experience of consistent patterns of experience.

On the “active” side, which is more clearly recognized as “habit,” proficiency in action execution also comes via prediction, but this time, instead of predicting how the distal structure of the world, we predict the same world we “self-fulfill,” as we act. Moving in the world feels like something to us (proprioception), and as we repeat activities, we become proficient in predicting the very sensory stimulation that we generate via our actions. The two sides of the double-law, which show up in contemporary predictive cognitive science as the difference between “perceptual” and “active” inference (Pezzulo et al., 2015; Wiese & Metzinger, 2017), are thus built on the predictive capacities of habits. This was something that was anticipated by Ravaisson when he noted that

[A] sort of obscure activity that increasingly anticipates both the impression of external objects in sensibility and the will in activity. In activity this reproduces the action itself; in sensibility it does not reproduce the sensation, the passion…but class for it, invokes it; in a certain sense it implores the sensation (Ravaisson 2008: 51).

Habit is thus the confluence of what has been called perceptual inference (predicting incoming signals by tuning a generative model of their causes) and active inference (self-fulfilling incoming signals via action so that they conform to the model that already exist), in other words, prediction as it facilitates our engaged coping with the world, is the nature of habit. More accurately, to the extent that we can predict the world, we do so via habit.

References

Carlisle, C. (2010). Between Freedom and Necessity: Félix Ravaisson on Habit and the Moral Life. Inquiry: A Journal of Medical Care Organization, Provision, and Financing, 53(2), 123–145.

Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press.

de Biran, P. M. (1970). The Influence of Habit on the Faculty of Thinking. Greenwood.

Emirbayer, M., & Mische, A. (1998). What is agency? The American Journal of Sociology, 103(4), 962–1023.

Hohwy, J. (2013). The Predictive Mind. Oxford University Press.

Pezzulo, G., Rigoli, F., & Friston, K. (2015). Active Inference, homeostatic regulation and adaptive behavioural control. Progress in Neurobiology, 134, 17–35.

Ravaisson, F. (2008). Of Habit. Bloomsbury Publishing.

Silver, D. (2011). The moodiness of action. Sociological Theory, 29(3), 199–222.

Sinclair, M. (2018). Habit and time in nineteenth-century French philosophy: Albert Lemoine between Bergson and Ravaisson. British Journal for the History of Philosophy: BJHP: The Journal of the British Society for the History of Philosophy, 26(1), 131–153.

Sinclair, M. (2019). Being Inclined: Félix Ravaisson’s Philosophy of Habit. Oxford University Press.

Wegner, D. M. (2002). The Illusion of Conscious Will. MIT Press.

Wiese, W., & Metzinger, T. (2017). Vanilla PP for Philosophers: A Primer on Predictive Processing. In T. Metzinger & W. Wiese (Eds.), Philosophy and Predictive Processing.

Explaining social phenomena by multilevel mechanisms

Four questions about multilevel mechanisms

In our previous post, we discussed mechanistic philosophy of science and its contribution to the cognitive social sciences. In this blog post, we will discuss three case studies of research programs at the interface of the cognitive sciences and the social sciences. In our cases, we apply mechanistic philosophy of science to make sense of the epistemological, ontological, and methodological aspects of the cognitive social sciences. Our case studies deal with the phenomena of social coordination, transactive memory, and ethnicity.

In our work, we have drawn on Stuart Glennan’s minimal account of mechanisms, according to which a mechanism for a phenomenon “consists of entities (or parts) whose activities and interactions are organized so as to be responsible for the phenomenon” (Glennan 2017: 17). We understand entities and activities liberally so as to accommodate the highly diverse sets of entities that are studied in the cognitive social sciences, from physically grounded mental representations to material artifacts and entire social systems. In our article, we make use of the following four questions drawn from William Bechtel’s (2009) work to assess the adequacy and comprehensiveness of mechanistic explanations:

What is the phenomenon to be explained (‘looking at’)?
What are the relevant entities and their activities (‘looking down’)?
What are the organization and interactions of these entities and activities through which they contribute to the phenomenon (‘looking around’)?
What is the environment in which the mechanism is situated, and how does it affect its functioning (‘looking up’)?

The visual metaphors of looking at the phenomenon to be explained, looking down at the entities and activities that underlie the phenomenon, looking around at the ways in which these entities and activities are organized, and looking up at the environment in which the mechanism operates, are intended to emphasize that mechanistic explanations are not strongly reductive or “bottom-up” explanations. Rather, multilevel mechanistic explanations can bring together more “bottom-up” perspectives from the cognitive sciences with more “top-down” perspectives from the social sciences in order to provide integrated explanations of complex social phenomena. In the following, we will illustrate how we have used mechanistic philosophy of science in our case studies and what we have learned from them.

Social Coordination

Interpersonal social coordination has been studied during recent decades in many different scientific disciplines, from developmental psychology (e.g., Carpenter&Svetlova 2016) to evolutionary anthropology (e.g., Tomasello et al. 2005) and cognitive science (e.g., Knoblich et al. 2011). However, despite their shared interests, there has so far been relatively limited interaction between different disciplinary research programs studying social coordination. In this case study, we argued that mechanistic philosophy of science can ground a feasible division of labor between researchers in different scientific disciplines studying social coordination.

In evolutionary anthropology and developmental psychology, one of the most important ideas that has gained considerable empirical support during recent decades is that human agents and our nearest primate relatives differ fundamentally in our dispositions to social coordination and cooperation: for example, chimpanzees rarely act together instrumentally in natural settings, and they are not motivated to engage in the types of social games and joint attention that human infants find intrinsically rewarding already at an early age (Warneken et al. 2006). Importantly, this does not seem to be due to a deficit in general intelligence since chimpanzees score as well as young human infants on tests of quantitative, spatial, and causal cognition (Herrmann et al. 2007). According to the shared intentionality -hypothesis of evolutionary anthropologist Michael Tomasello, this is because “human beings, and only human beings, are biologically adapted for participating in collaborative activities involving shared goals and socially coordinated action plans (joint intentions)” (Tomasello et al. 2005).

Given a basic capacity to engage in social coordination, one can raise the question of what types of cognitive mechanisms enable individuals to share mental states and act together with other individuals. To answer this question, we made use of the distinction between emergent and planned forms of coordination put forth by cognitive scientist Günther Knoblich and his collaborators. According to Knoblich et al. (2011: 62), in emergent coordination, “coordinated behavior occurs due to perception-action couplings that make multiple individuals act in similar ways… independent of any joint plans or common knowledge”. In planned coordination, ”agents’ behavior is driven by representations that specify the desired outcomes of joint action and the agent’s own part in achieving these outcomes.” Knoblich et al. (2011) discuss four different mechanisms for emergent coordination: entrainment, common object affordances, action simulation, and perception-action matching. While emergent coordination is explained primarily by sub-intentional mechanisms of action control (which space does not allow us to discuss in more detail here), planned coordination is explained by reference to explicit mental representations of a common goal, the other individuals in joint action, and/or the division of tasks between the participants.

In our article, we argued that cognitive scientists and social scientists answer different questions (see above) about mechanisms that bring about and sustain social coordination in different environments and over time. Thus they are in a position to make mutually interlocking yet irreducible contributions to a unified mechanistic theory of social coordination, although they may also sometimes reach results that challenge assumptions that are deeply ingrained in the other group of disciplines. For a more detailed discussion of how cognitive and social scientists can collaborate in explaining social coordination, we refer the reader to our article (Sarkia et al. 2020: 8-11).

Transactive Memory

Our second case study concerned the phenomenon of transactive memory, which has been studied in the fields of cognitive, organizational, and social psychology as well as in communication studies, information science, and management. The social psychologist Daniel Wegner and his colleagues (Wegner et al. 1985: 256) define transactive memory in terms of the following two components:

An organized store of knowledge that is contained entirely in the individual memory systems of the group members
A set of knowledge-relevant transactive processes that occur among group members.

They attribute transactive memory systems to organized groups insofar as these groups perform functionally equivalent roles in group-level information processing as individual memory mechanisms perform in individual cognition, i.e. (transactive) encoding, (transactive) storing, and (transactive) retrieving of information. For example, Wegner et al. (1985) found that close romantic couples responded to factual and opinion questions by using integrative strategies, such as interactive cueing in memory retrieval. Subsequent research on transactive memory systems has addressed small interaction groups, work teams, and organizations in addition to intimate couples (e.g., Ren & Argote 2011; Peltokorpi 2008). What is crucial for the development of a transactive memory system is that the group members have at least partially different domains of expertise and that the group members have learned about each other’s domains of expertise. If these two conditions are met, each group member can utilize the other group members’ domain-specific information in group-related cognitive tasks and transcend the limitations of their own internal memories.

In our article, we made use of the theory of transactive memory systems to argue that some cognitive mechanisms transcend the brains and bodies of individuals to the social and material environments that they inhabit. For example, in addition to brain-based memories, individual group members may also utilize material artifacts, such as notebooks, archives, and data files, as their memory stores. In addition, other members’ internal and external memory storages may in an extended sense be understood as part of the focal member’s external memory storages as long as she knows their domains of expertise and can communicate with them. Thus the theory of transactive memory can be understood as describing a socially distributed and extended cognitive system that goes beyond intra-cranial cognition (Hutchins 1995; Sutton et al. 2010). For a more detailed discussion of this thesis and its implications for interdisciplinary memory studies, we refer the reader to our article (Sarkia et al. 2011, 11-15).

Ethnicity

The sociologist Rogers Brubaker and his collaborators (Brubaker et al. 2004) has made use of theories in cognitive psychology and anthropology to challenge traditional approaches to ethnicity, nationhood, and race that view them as substantial groups or entities with clear boundaries, interests, and agency. Rather, he treats them as different ways of seeing the world, based on universal cognitive mechanisms, such as categorizing the world into ‘us’ and ‘them.’ Brubaker et al. (2004) also make use of the notions of cognitive schema and stereotype, defining stereotypes as “cognitive structures that contain knowledge, beliefs, and expectations about social groups” and schemas as “representations of knowledge and information-processing mechanisms” (DiMaggio 1997). For example, Brubaker et al. (2004, 44) discuss the process of ethnicization, where ”ethnic schemas become hyper-accessible and… crowd out other interpretive schemas.”

In our article, we made use of Brubaker’s approach to ethnicity to illustrate how cognitive accounts of social phenomena need to be supplemented by traditional social scientific research methods, such as ethnographic and survey methods when we seek to understand the broader social and cultural environment in which cognitive mechanisms operate. For example, in their case study of Cluj, a Romanian town with a significant Hungarian minority, Brubaker et al. (2006) found that while public discourse was filled with ethnic rhetoric, ethnic tension was surprisingly scarce in everyday life. By collecting data with interviews, participant observation, and group discussions, they were able to identify cues in various situations that turned a unique person into a representative of an ethnic group. Importantly, this result could not be achieved simply by studying the universal cognitive mechanisms of stereotypes, schemas, and categorization, since these mechanisms serve merely as the vehicles of ethnic representations, and they do not teach us about the culture-specific contents that these vehicles carry. We refer the reader to our article for more discussion of the complementarity of social scientific and cognitive approaches to ethnicity (Sarkia et al. 2020, 15-17).

References

Bechtel W (2009) “Looking down, around, and up: mechanistic explanation in psychology.” Philosophical Psychology 22(5): 543–564.

Brubaker R, Loveman M and Stamatov P (2004) “Ethnicity as cognition.” Theory and Society 33(1): 31–64.

Brubaker R, Feischmidt M, Fox J, Grancea L (2006) Nationalist Politics and Everyday Ethnicity in a Transylvanian Town. Princeton: Princeton University Press.

Carpenter M, Svetlova M (2016) “Social development.” In: Hopkins B, Geangu E, Linkenauer S (eds) Cambridge Encyclopedia of Child Development. Cambridge: Cambridge University Press, 415–423.

DiMaggio P (1997) “Culture and cognition.” Annual Review of Sociology 23: 263-287.

Herrmann E, Call J, Hernandez-Loreda, M, Hare B, and Tomasello, M (2007). “Humans have evolved specialized skills of social cognition: the cultural intelligence hypothesis.” Science 317: 1360-1366.

Hutchins E (1995) Cognition in the wild. Cambridge (MA): MIT Press.

Peltokorpi V (2008). Transactive memory systems. Review of General Psychology 12(4): 378–394.

Ren Y and Argote L (2011) “Transactive memory systems 1985–2010: An integrative framework of key dimensions, antecedents, and consequences.” The Academy of Management Annals 5(1): 189–229.

Sarkia M, Kaidesoja T, and Hyyryläinen (2020). “Mechanistic explanations in the cognitive social sciences: lessons from three case studies.” Social Science Information. Online first (open access). https://doi.org/10.1177%2F0539018420968742

Sutton J, Harris C.B., Keil P.G. and Barnier A.J. 2010. “The psychology of memory, extended cognition and socially distributed remembering.” Phenomenology and the Cognitive Sciences 9(4), pp. 521-560.

Tomasello M, Carpenter M, Call J, et al. (2005) “Understanding and sharing intentions: The origins of cultural cognition.” Behavioral and Brain Sciences 28: 675–691.

Warneken F, Chen F, Tomasello M (2006) “Cooperative activities in young children and chimpanzees.” Child Development 77(3): 640–663.

Wegner DM, Giuliano T and Hertel P (1985) “Cognitive interdependence in close relationships.” In: Ickes WJ (ed) Compatible and Incompatible Relationships. New York: Springer, pp. 253–276.

Causal mechanisms in the cognitive social sciences

The social sciences and the cognitive sciences have grown closer together during recent decades. This is manifested in the emergence and expansion of new research fields, such as social cognitive neuroscience (Cacioppo et al. 2012; Lieberman 2017), cognitive sociology (Brekhus & Ignatow 2019), behavioral economics (Dhami 2016), and new approaches in cognitive anthropology (Bloch 2012; Hutchins 1995; Sperber 1996). However, increasing interactions between the cognitive and social sciences also raise many pressing philosophical and methodological issues about interdisciplinary integration and division of labor between disciplines. In our recent article (Sarkia, Kaidesoja & Hyyryläinen 2020), we argue that mechanistic philosophy of science can contribute to analyzing these challenges and responding to them.

According to mechanistic philosophy of science (hereafter: MPS), the primary way in which scientists explain complex cognitive and social phenomena is by describing causal mechanisms that produce, underlie, or maintain these phenomena (e.g. Bechtel 2008; Glennan 2017; Hedström & Ylikoski 2010). Commonly cited examples of semi-general social mechanisms include those that generate self-fulfilling prophecies, cumulative advantage, residential segregation, collective action, and diffusion patterns in social networks. Cognitive and neural mechanisms addressed in the cognitive sciences include those underlying perceptual processes, memory functions, learning, imagination, and social cognition.

In this post, we take a closer look at causal mechanisms and mechanistic explanations. We also indicate some ways in which MPS could help to bridge the gap between the social and the cognitive sciences. The text partially draws on our article that provides a more detailed account of mechanistic explanations in the cognitive social sciences (Sarkia, Kaidesoja & Hyyryläinen 2020: 3-8).

Mechanisms

A ‘minimal’ account of mechanisms says that a mechanism for a phenomenon “consists of entities (or parts) whose activities and interactions are organized so as to be responsible for the phenomenon” (Glennan 2017: 17). Entities are particular things (in a broad sense) in the world and activities always take place in some entity. The entities that are studied in different sciences are highly diverse, ranging from molecules to brains and complex social systems. Entities may engage in activities either by themselves or in concert with other entities. When the activities of two or more entities influence each other, they interact. In a mechanism that is responsible for some phenomenon, its constituent entities and activities, as well as their interactions, are organized in a way that allows them to produce, maintain or underlie the phenomenon, meaning that there are specific constitutive and causal relations between these constituent entities and activities. This minimal account of mechanisms makes clear that mechanisms are different from universal laws, correlations between variables (or other empirical regularities), and functions that items may perform in some larger system. Advocates of MPS have also provided accounts of mechanisms that are more specific, but most of them are compatible with the minimal account (e.g. Glennan & Illari 2018).

MPS regards mechanisms as hierarchical in the sense that lower-level mechanisms operate as parts of higher-level mechanisms (e.g. Craver & Darden 2013; Glennan 2017). When scientists investigate a mechanism that is responsible for a specific phenomenon, they commonly assume that there are underlying mechanisms that allow the constituent entities of the mechanism to engage in the activities that they engage in. Conversely, a mechanism identified at a lower level of mechanistic organization is typically embedded in some broader (or higher-level) mechanism that affects its functioning. For example, a mechanism underlying the working memory of a particular person may operate as a part of the social mechanism of collaborative learning in which the person is engaged in a common learning task with her classmates. Social and cognitive scientists often implicitly or explicitly attribute different types of cognitive capacities to people, such as the capacities to act intentionally, to communicate using spoken or written language, and to remember things from the past. As Stuart Glennan (2017: 51–52) argues, the capacities of complex entities are mechanism-dependent in the sense that the organized interactions of their parts are responsible for the capacities of the whole entity, which may manifest themselves only in suitable environments. For example, the capacity for speech is dependent on the organized interactions of neural mechanisms and manifested in embodied communicative interactions with other people.

According to MPS, mechanisms are identified on the basis of the phenomena that they contribute to (e.g. Craver & Darden 2013; Hedström & Ylikoski 2010; Glennan 2017). For example, cognitive neuroscientists investigate the neural mechanisms underlying working memory and visual perception (Bechtel, 2008), while social scientists study the social mechanisms of self-fulfilling prophecy and urban segregation (Hedström, 2005). They both use empirically established phenomena to delimit the boundaries of the mechanism under investigation and to identify the entities and activities that are relevant for explaining the phenomenon in question.

When they study highly complex systems, such as biological organisms or social groups, scientists may also get different mechanistic decompositions of the same system when they focus on different phenomena in the system (Glennan 2017: 37–38). But once they have identified a phenomenon in a system, the boundaries of the mechanism that is responsible for it are determinate and do not depend on the ways the mechanism is represented. An important implication of this is that mechanistic levels are always relative to some phenomenon of interest, meaning that there are no global levels of mechanisms. From this, it follows that cognitive social scientists should be cautious regarding the methodological value of highly abstract mechanism types, such as ‘biological mechanism’, ‘psychological mechanism’ and ‘social mechanism’ since they tend to refer to heterogeneous arrays of mechanisms rather than to fixed ‘ontological levels of reality’.

Mechanistic Explanations

While mechanisms are always particular and spatiotemporally local, cognitive and social scientists are interested in making generalizations about them and classifying them into kinds. According to MPS, scientists achieve generality by constructing models about classes of particular mechanisms. In scientific practice, mechanistic models may take many different forms, such as qualitative descriptions, diagrams, equations, or computational simulations. What they share in common is that they can be used to ‘describe (in some degree and some respect) the [target] mechanism that is responsible for some phenomenon’ (Glennan 2017: 66). An important way to construct general models is by abstracting away from the details of particular mechanisms and idealizing some of their features. For example, many models of social mechanisms not only abstract away from most neural and cognitive mechanisms that underlie the interactions of individual actors but may also include idealized descriptions of the cognitive capacities of actual human beings (cf. Hedström, 2005; Hedström & Ylikoski 2010). Abstractions omit details regarding the target mechanism while idealizations distort some features of the target mechanism (Craver & Darden 2013: 33–34, 94; Glennan 2017: 73–74). There is no general criterion regarding the acceptability of abstractions and idealizations in a mechanistic model – rather, the appropriateness of particular abstractions and idealizations should be decided in a case-by-case manner depending on the epistemic aims of the researcher (Craver & Kaplan 2018; Glennan 2017).

In MPS, scientific explanations are understood in terms of mechanistic models that scientists use –in combination with other relevant explanatory factors – to represent those mechanisms that underlie, maintain or produce the phenomenon that they aim to explain (e.g. Bechtel 2008; Craver & Darden 2013; Glennan 2017). Mechanistic explanations may unify phenomena that were earlier regarded as unconnected by revealing that their underlying mechanisms are similar. Mechanistic explanations may also split phenomena that were earlier regarded as similar by revealing that their underlying mechanisms are different.

In the context of the cognitive social sciences, some researchers have recognized the identification of cognitive mechanisms underlying social phenomena as a central argument for the cognitive social sciences (e.g. Sun 2017; Thagard 2019), while others have argued in favor of greater unification (e.g. Gintis 2007), complementarity (e.g. Zerubavel 1997) or mutual constraints (e.g. Bloch 2012) between the cognitive and social sciences without appeal to mechanistic philosophy of science. We have discussed different arguments for the cognitive social sciences in more detail in an earlier article (Kaidesoja et al. 2019) and a blog post that was based on it. However, when evaluating mechanistic explanations for social phenomena, it is important to recognize that such explanations do not reduce the phenomena to be explained to some lower level. Rather, they help us to understand how the phenomena to be explained arise from the organized interactions of its constituent entities and activities in a specific environment. This means that mechanistic explanations often cite mechanisms at many different levels in a local mechanistic hierarchy.

Some critics of MPS have claimed that advocates of this view assume that more detailed mechanistic explanations are always better (e.g. Batterman & Rice 2014), although the latter have explicitly distanced their views from this idea (e.g. Glennan 2017; Craver & Kaplan 2018). Even if it is clear that a mechanistic explanation should describe some entities and activities that contribute to the phenomenon to be explained, mechanistic explanations may vary with respect to their completeness, and the epistemic purposes of researchers should be taken into account when assessing the relevance of adding more detail to a mechanistic model. Accordingly, in their well-known article on causal mechanisms in the social sciences, Peter Hedström and Petri Ylikoski (2010: 60) conclude that ‘only those aspects of cognition that are relevant for the explanatory task at hand should be included in the explanation, and the explanatory task thus determines how rich the psychological assumptions must be’. Cognitive explanations of social phenomena may accordingly involve various degrees of realism and complexity, and more detailed multi-level explanations are not automatically more satisfactory than explanations that focus on a more straightforward or selective subset of causes.

Conclusion

This brief account of causal mechanisms and mechanistic explanations already provides some ideas on how to integrate the social sciences with the cognitive sciences. In the simplest case, mechanisms studied in the cognitive and social sciences can be organized in a hierarchical manner such that cognitive scientists model those cognitive and neural mechanisms that directly underlie those cognitive capacities and activities of social actors that are assumed in social scientists’ models about social mechanisms. However, few mechanistic models in the cognitive and social sciences can be organized into vertical relations of this kind. It is often the case, for example, that cognitive scientific and social scientific models address partially overlapping phenomena in different spatiotemporal scales by using different conceptual frameworks and research methods (e.g. Bloch 2012; Lizardo et al 2020; Turner 2018). This means that there are still significant conceptual gaps and methodological discrepancies that cognitive social scientists need to address in their explanatory practices. In our paper, we used MPS to address some of these difficulties and applied it in three case studies about the cognitive social sciences. In a follow-up post, we discuss our case studies and their lessons.

References

Batterman, RW, and Rice C (2014) “Minimal model explanations.” Philosophy of Science 81(3): 349–76.

Bechtel W (2008) Mental Mechanisms: Philosophical Perspectives on Cognitive Neuroscience. Routledge: London.

Bloch M (2012) Anthropology and the Cognitive Challenge. Cambridge: Cambridge University Press.

Brekhus W and Ignatow G (eds) (2019) Oxford Handbook of Cognitive Sociology. Oxford: Oxford University Press.

Cacioppo J, Berntson G and Decety J (2012) “A history of social neuroscience.” In: Kruglanski A and Stroebe W (eds) Handbook of the History of Social Psychology. New York: Psychology Press, pp.123-136.

Craver C and Darden L (2013) In Search of Mechanisms: Discoveries Across the Life Sciences. Chicago: University of Chicago Press.

Craver C and Kaplan D (2018) “Are more details better? On the norms of completeness for mechanistic explanations.” The British Journal for the Philosophy of Science, 1(71): 287–319

Dhami S (2016) The Foundations of Behavioral Economic Analysis. Oxford University Press.

Gintis H. (2007) A framework for the unification of the behavioral sciences. Behavioral and Brain Sciences, 30: 1–16.

Glennan S (2017) The New Mechanical Philosophy. Oxford: Oxford University Press.

Glennan S and Illari P (eds) (2018) The Routledge Handbook of Mechanisms and Mechanical Philosophy. London: Routledge.

Hedström, P (2005) Dissecting the Social: On the Principles of Analytical Sociology. Cambridge: Cambridge University Press.

Hedström P and Ylikoski P (2010) “Causal mechanisms in the social sciences.” Annual Reviews in Sociology 39: 46-67.

Hutchins E (1995) Cognition in the wild. Cambridge: MIT Press.

Kaidesoja T, Sarkia M and Hyyryläinen M (2019) “Arguments for the cognitive social sciences.” Journal for the Theory of Social Behavior 49(4):1-16. https://onlinelibrary.wiley.com/doi/abs/10.1111/jtsb.12226

Lieberman M (2017) “Social cognitive neuroscience: A review of core processes.” Annual Review of Psychology 58: 259–289.

Lizardo O, Sepulvado B, Stoltz D and Taylor M (2020) “What can cognitive neuroscience do for cultural sociology.” American Journal of Cultural Sociology 8: 3–28.

Sarkia M, Kaidesoja T and Hyyryläinen M (2020) “Mechanistic explanations in the cognitive social sciences: Lessons from three case studies.” Social Science Information. https://doi.org/10.1177%2F0539018420968742

Sperber D (1996) Explaining Culture: a Naturalistic Approach. Oxford: Blackwell.

Sun R (2012) “Prolegomena to the cognitive social sciences.” In R. Sun (ed) Grounding Social Sciences in Cognitive Sciences. Cambridge (MA): MIT Press, pp. 3–32.

Thagard P (2019) Mind-Society: From Brains to Social Sciences and Professions. Oxford: Oxford University Press.

Turner SP (2018) Cognitive Science and the Social. London: Routledge.

Zerubavel, E (1997) Social Mindscapes: An Invitation to Cognitive Sociology. Cambridge, MA: Harvard University Press.

Did John Dewey Put Prediction into Action?

Prediction does not appear, at first, to be something that a sociologist, or really any analyst of anything, can safely ascribe to those (or that) which they are studying without running afoul of about a thousand different stringent rules that define how probability can be used for the purposes of generating knowledge. If we follow the likes of Ian Hacking (1975) and Lorraine Daston (1988) (among others), then “modern fact-making” has a lot to do with ways of using probability, especially for the purposes of making predictions. To the degree that this transforms probability into prediction, as referring to the epistemic practices that analysts use to generate a knowledge claim, this usage actually places limits on what probability can mean, how prediction can be used and where we might find it. If we don’t have certain epistemic practices (e.g. a nice regression analysis) then we can’t say that prediction is occurring anywhere if we are not doing it ourselves.

As Hacking and Daston indicate, however, for probability to be limited almost entirely to epistemic practices in this sense would appear strange to those who can stake any sort of claim to having “discovered” probability, especially Blaise Pascal. He, for one, did not understand probability to be limited to efforts at making predictions for the purposes of knowledge. For Pascal, probability had direct analogues in lived experience (without calculation) in the form of senses of risk and high stakes, and the perceived fairness of outcome, particularly in games of chance. If this seems unusual to us now, given the strictures we place on probability and prediction, these points are far less unusual for what is fast appearing as a major paradigm in cognitive science, namely predictive processing (see Clark 2013; Friston 2009; Wiese and Metzinger 2017; Williams 2018; Hohwy 2020).

To put it simply, predictive processing makes prediction the primary function of the brain. The brain evolved to allow for the optimal form of engagement with a contingent and probabilistic environment that is never in a steady state. Given that our grey matter is locked away inside a thick layer of protective bone (e.g. the skull), it has no direct way of perceiving or “understanding” what is coming at it from the outside world. What it does have are the senses, which themselves evolved to gather information about that environment. Predictive processing says, in essence, that the brain can have “knowledge” of its environment by building the equivalent of a model and using it to constantly generating predictions about what the incoming sensory information could be. This works in a continuous way, both at the level of the neuron and synapse, and at the level of the whole organism. The brain does not “represent” what it is dealing with, then, but it uses associations, co-occurrences, tendencies and rhythms to predict what it is dealing with.

All of this is contingent on making the equivalent of constant, future-oriented but past-deriving, best guesses. When those guesses are wrong, this generates error, which forms the content of our perceptions. In other words, what we perceive and consciously attend to is the leftover error of our generative models and their predictions of our sensory input. When those guesses are right, by contrast, we don’t have perceptual content because there is no error. The generative models we build are themselves multi-tiered, and the predictions they make work at several different levels of composition. A full explanation of predictive processing far exceeds the limits of this post. But this, in particular, is worth mentioning because it means that a generative model is not static or unchanging. Quite the contrary, generative models constantly change (at some compositional level) in order to better ensure prediction error minimization.

Some of these points will probably not sound that unusual. The relationship between minimized perceptual content and action is commonly referred to in discussions of embodiment and moral intuition, for example. What probably sounds very unusual, however, is the central role given here to prediction.

As mentioned, prediction has been essentially cordoned off in the protected sphere of knowledge, to be used only by specialists wielding specialist tools and training. While it can be done by the folk, we (the analysts) love to point out how they do it poorly. On the off chance they happen to predict correctly (e.g. gambling on the long shot), this is celebrated as the exception that proves the rule. After all, the folk do not have our epistemic practices or training. All they have is their (subjective) experience and biases. In fact, brandishing those presumably bad at predicting by those with increasingly sophisticated techniques to make predictions on increasingly large datasets has become par for the course in the era of “analytics” (Hofman, Sharma and Watts 2017), and this particular symbolic power is now wielded quite overtly in a variety of fields (like baseball). Thus, to take prediction away from action could have, all along, been just another way of saying that because we (the analysts) predict and they (the folk) do not predict or do so poorly, they need us.

But is this the case? Predictive processing poses a serious question to this assumption and, with it, the role that prediction plays in making sociological knowledge different from folk knowledge. There is also a bit of history worth mentioning. The assumption that prediction plays only a negligible part in action, while other things like values and beliefs play a big part, comes from Talcott Parsons, who explicitly set out to marginalize prediction (1937: 64). Sociologists are rightfully in the mood of poo-pooing Parsons and have been for quite some time; but any proposal to put prediction into action remains just as heretical today as it did to Parsons in the 1930s. As one of his major points about action, the presumption that prediction can play no direct or significant role in action has still not been revisited let alone revised.

The purpose of this post is simply to sketch out the suggestion that we can even do this (e.g. put prediction into action) without falling over our feet and retreating sheepishly to the safety of the domain the Parsons carved out for us should we ever wish to talk about “action” again. Far be it from me to attempt to do this on my own. So for the purposes of illustration, a few pages from John Dewey’s Logic: The Theory of Inquiry (1938: 101-116) (and few from Human Nature and Conduct [1922]) will be enlisted for the task. I will argue that, in these pages, which are themselves famous because in them Dewey gives specific proposals about the process or stages of inquiry, Dewey does put prediction into action, and he does so in a way that does not seem that controversial; though, for any legitimate contemporary meaning of “prediction,” these are heretical claims.

For Dewey, in contrast to Parsons, the action situation is not neatly parsed into the “objective state of affairs” that could be described with scientific precision by an external observer (and for which prediction is appropriate) and the “subjective point of view” of the actor (for which, by implication, prediction does not apply, lest we “squeeze out” the creative, voluntaristic element). Instead, the “state of affairs” is, according to Dewey (1922, p. 100ff), irreducibly composed of an entanglement of both objective and subjective elements. The very act of perception of a given state of affairs on the part of the actor introduces such a subjective element (for Parsons perception was not necessarily part of the subjective element of the action schema).

Perception is not just purely spectatorial or contemplationist, then, but serves as the “initial stage” in a dynamic action cycle. Perception is for something, and this something is anticipation and prediction. Thus, “the terminal outcome when anticipated (as it is when a moving cause of affairs is perceived) becomes an end-in-view, an aim, purpose, a prediction usable as a plan in shaping the course of events” (Dewey 1922:101, italics added). In a stronger sense, for Dewey perceptions are predictions, which in their turn are ends-in-view. Perceptions are “projections of possible consequences; they are ends-in-view. The in-viewness of ends is as much conditioning by antecedent natural conditions as is perception of contemporary objets external to the organism, trees and stones or whatever” (102).

For Dewey (1938), this can extend even further into what arguably remains his most influential contribution to pragmatist thought: the process of inquiry, as it “enters into every aspect of every area” of life (101). Inquiry, as Dewey defines it, is the “controlled or directed transformation of an indeterminate situation into one that is so determinate in its constituent distinctions and relations as to convert the elements of the original situation into a unified whole” (104-105). This filters into all subsequent understandings of pragmatist problem-solving.

The “indeterminate situation” (105) that provides antecedent conditions for inquiry is constituted by doubt, but this is not a purely subjective state (“in us”). Doubt refers to our placement in a situation that is doubtful because we cannot respond to it as we are accustomed: “the particular quality of what pervades the given materials, constituting them a situation … is a unique doubtfulness which makes that situation to be just and only the situation it is” (105). Specifically this means that we cannot form ends-in-view with respect to the situation, though we can “[respond] to it … [in] blind and wild overt activities.” As Dewey stresses, “it is the situation that has these traits,” which means that we are simply a part of the situation in being doubtful; one part of the total configuration. To simply “change our mind” with respect to the doubtful situation is hardly enough to change it, though with any indeterminate situation, we might respond by carrying through a “withdrawal from reality.” The only thing that will really be effective, however, is what Dewey calls a “restoration of integration” in which the situation changes as our situation within it changes (e.g. as we change) (106).

Underlying Dewey’s proposals, then, is a kind of cognitive mechanism, which he does not label outright, but which, likewise, rests on prediction, and on which the stages of inquiry itself appear to rest. For Dewey (107-108), it is possible to remain in the doubtful situation forever, particularly should you find an effective means of “withdrawing from reality.” The next stage in the process of inquiry will only occur through a change in “cognitive operations,” specifically what Dewey labels “the institution of the problem … The first result of evocation of inquiry is that the situation is taken, adjudged, to be problematic. To see that a situation requires inquiry is the initial step in inquiry” (107). But to take this step, as Dewey implies, requires a change in the manner of prediction, and in a not dissimilar sense as a roughly equivalent mechanism identified by predictive processing.

If the indeterminate situation does not allow for perceptions as “ends-in-view,” then in the problematic situation the actor (e.g. “the interpretant”) changes because, in the situation, she is now characterized by an explicit representation: “without a problem, there is blind grasping in the dark.” This representation is needed as a change in cognition, but only as a mediating and not a permanent state. But the constant in this process, that allows representation to appear now explicitly and then only to disappear later on, can only be successive forms of prediction that, in Dewey’s terms, is trying to obtain an end-in-view. In other words, the explicit representation of “problem” itself presupposes a prediction about error. More generally, we are part of a problematic situation because we predict that it should go one way and it does not, and then we anticipate what would be required to minimize that error, which then forms the basis for future action. In almost a directly analogous sense, predictive processing refers to this as “active inference.”

Hence, what follows this (“the determination of a problem-situation”) is subsequently characterized by the generation of “ideas” as part of the inherently progressive nature of inquiry along the lines of continuous prediction or forward-searching (e.g. guessing): “The statement of a problematic situation in terms of a problem has no meaning save as the problem instituted has, in the very terms of its statement, reference to a possible solution” (108). Put differently, the one (problem) never occurs without the other (solution); we actively infer solutions because we have problems. Dewey (110-111) uses this to critique all prior conceptions of “ideas” in a western philosophical tradition (empiricists, rationalists and Kantians) for not seeing how perceptions and ideas function correlatively rather than separately:

Observations of facts and suggested meanings or ideas arise and develop in correspondence with each other. The more the facts of the case come to light in consequence of their being subjected to observation, the clearer and more pertinent become the conceptions of the way the problem constituted by these facts is to be dealt with. On the other side, the clearer the idea, the more definite … become the operations of observation and of execution that must be performed in order to resolve the situation (109).

Ideas are not removed from the situation, or entirely defined by the situation. Rather, the most important thing about them is that they have a direction in relation to the situation. But this only works if they suggest a forward-facing (temporally speaking) cognitive mechanism, which again seems perfectly analogous to a predictive function that is trying (slowly) to minimize error. Dewey seeks to redeem the role of “suggestions” (which have “received scant courtesy in logical theory”) by giving them not the diminished importance of half-completed ideas, but elevating them to “the primary stuff of logical ideas.” In this sense, suggestions demonstrate how “perceptual and conceptual materials are instituted in functional correlativity to each other in such a manner that the former locates and describes the problem which the latter represents a method of solution” (111; emphasis added).

To “reason,” then, means to examine the meaning of ideas according to their simultaneous statement of problem and solution (e.g. “relationally”). For Dewey, this process involves “operating with symbols (constituting propositions)… in the sense of ratiocination and rational discourse.” If a suggested meaning is “immediately accepted,” then the inquiry will end prematurely. Full reasoning consists of a kind of “check upon immediate acceptance [as] the examination of … the meaning in question” according to what it “implies in relation to other meanings in the system of which it is a member” (111). By “meaning” Dewey refers to symbols in a semiotic sense or the connection of sign and object in a non-problematic or habitual way. This therefore opens those habitual associations up to transformation as the situation becomes more determinate. Dewey also emphasizes how symbols perform the semiotic function of “fact-meanings.” The process of inquiry subjects these connections to “ideas [as] operational in that they instigate and direct further operations of observation; they are proposals and plans for acting upon existing conditions to bring new facts to light and to organize all the selected facts into a coherent whole” (112-113). The process remains forward-facing, which means that there can be “trial facts” that can be taken on-board with a certain provisionality: “they are tested and ‘proved’ with respect to their evidential function.” Ideas and facts, then, become “operative” in the process of inquiry (problem-solving) “to the degree in which they are connected with experiment” (114). Again, all of this presupposes that forward momentum, or searching, appears to be fueled by advancing and constant prediction.

Thus, for Dewey, the transformation of the situation into “determinate” involves a change of “symbols” in the form of habitual associations (sign to object) which themselves always remain provisional and never fully determinate (114-115). This is what alters our “self” (interpretant) within the situation as no longer in a doubtful state, and replaces this with what we might call a “confident” state as signifying a kind of assurance of action in relation to the situation.

Thus, having passed through the stages of inquiry, and with new habitual associations, we are now predicting it well within the continuous flow of action. In Dewey’s terms, problem and solution effectively merge at the end of inquiry, and the forward-facing search ends. But we can translate the folk terms that Dewey uses here almost directly into the more technical terms that form the basis of predictive processing: the problem or trial-situation ends with the erasure of prediction error by a change in the generative model, such that the tiered coding of sensory input will generate the perceptions that the generative model expects. X is now Y in a non-problematic way, which for Dewey means that it becomes a “symbol” as a connection that is now habitual (see also Peirce CP 2: 234). Inquiry in “common sense” and inquiry in science are not different, according to Dewey, they simply involve differences in problems. For common sense, problems appear from symbols as the habitual culture of groups (115-116).

This can lead us to make an even more radical claim: prediction in action and prediction in sociology are also not different; they simply involve differences in problems between those that occur in the continuous course of action, and those that are deliberately manufactured for the purposes of staging trials and leveraging them in order to make knowledge claims. Shared generative models also appear among actually-existing groups that make similar predictions, perceive similar things based on similar error, make similar active inferences, and therefore “solve problems” in ways that have a family resemblance.

It seems then, without too much presumptuousness, we can take Dewey’s original definition of inquiry and retranslate it into its implied cognitive terms:

The controlled or directed transformation of an indeterminate situation into one that is so determinate in its constituent distinctions and relations as to convert the elements of the original situation into a unified whole (Dewey 1938: 104-105).

We can translate this into a general statement about problem-solving as follows

The higher order transformation of a situation with lots of prediction error into a generative model that is able to convert the elements of the original situation into a predictable whole.

A follow-up post will discuss the broader significance of this translation in relation to pragmatist theories of action.

References

Clark, Andy. 2013. “Whatever next? Predictive Brains, Situated Agents, and the Future of Cognitive Science.” The Behavioral and Brain Sciences 36(3):181–204.

Daston, Lorraine. 1988. Classical Probability in the Enlightenment. Princeton University Press.

Dewey, John. 1938. Logic: The Theory of Inquiry. New York: Holt, Reinhart and Winston.

Dewey, John. 1922. Human Nature and Conduct. New York: Henry Holt.

Friston, Karl. 2009. “The Free-Energy Principle: A Rough Guide to the Brain?” Trends in Cognitive Sciences 13:293–301.

Hacking, Ian. 1975. The Emergence of Probability: A Philosophical Study of Early Ideas about Probability, Induction and Statistical Inference. Cambridge University Press.

Hofman, Jake M., Amit Sharma, and Duncan J. Watts. 2017. “Prediction and Explanation in Social Systems.” Science 355(6324):486–88.

Hohwy, Jakob. 2020. “New directions in predictive processing.” Mind and Language 35: 209-223.

Parsons, Talcott. 1937. The Structure of Social Action. New York: Free Press.

Wiese, Wanja, and Thomas Metzinger. 2017. “Vanilla PP for Philosophers: A Primer on Predictive Processing.” in Philosophy and Predictive Processing.

Williams, Daniel. 2018. “Pragmatism and the Predictive Mind.” Phenomenology and the Cognitive Sciences 17:835–59.

The Cognitive Hesitation: or, CSS’s Sociological Predecessor

Simmel is widely considered to be the seminal figure from the classical sociological tradition on social network analysis. As certain principles and tools of network analysis have been transposed to empirical domains beyond their conventional home, Simmel has also become the classical predecessor for formal sociology, giving license to the effort and providing a host of formal techniques with which to pursue the work (Erikson 2013; Silver and Lee 2012). As Silver and Brocic (2019) argue, part of the appeal of Simmel’s “form” is its pragmatic utility and adaptability. Simmel demonstrates this in applying different versions of form to different empirical objects ( e.g. “the stranger” versus “exchange”). This suggests that we need not make much headway on deciphering what “form” actually is and still practice a formal sociology.

Though it may not seem like it, these recent efforts at formal sociology find their heritage in a sometimes rancorous debate etched deeply into Simmel’s cross-Atlantic translation into American sociology (and therefore not insignificant on shaping cross-field perceptions of sociology as “science”). Historically, this has found proponents of a middle-range application of form set against those who appeal to a more diffuse concern with the status of form. The debate has proven contentious enough, including at least one occasion of translation/retranslation of terminology from Simmel’s work. Robert Merton retranslated the German term ubersehbar to mean “visible to” (in the sentence from “The Nobility” [or Aristocracy”] discussion in Soziologie: “If it is to be effective as a whole, the aristocratic group must be “visible to” [ubersehbar] every single member of it. Each element must be personally acquainted with every other”) instead of what Kurt Wolff had originally translated as “surveyable by.” For various reasons, “visible to” carried far less of a “phenomenological penumbra” and fit with Merton’s interest (e.g. disciplinary position-taking) in structure, but arguably did not match Simmel’s own interest in finding the “vital conditions of an aristocracy” (see Jaworski 1990).

More recently, a kind of detente has emerged between the two sides. To the degree that there is any concern for the status of “form” itself, formal sociology has taken on board what is arguably the most thoroughgoing defense of Simmel’s “phenomenology” to date: the philosopher Gary Backhaus’ 1999 argument for Simmel’s “eidetic social science.” Backhaus reads Simmel with the help of Edmund Husserl, the founder of phenomenology, and therefore reads him against the grain of what the philosophically-minded had conventionally read as Simmel’s more straightforward neo-Kantianism. In part, this detente with phenomenology has been done because Backhaus made it easy to do. His reading does not require that formal sociology do anything that would deviate from network analysis’ own bracketing of the content of social ties from the formal pattern of social ties. His reading of Simmel also remains compatible with a pluralist/pragmatist application of form.

The purpose of this post is threefold: (1) to question that the status of Simmel’s “form” is philosophical and therefore capable of being resolved into either a phenomenology or neo-Kantianism; (2) to situate Simmel as part of a lost 19th century interscience (volkerpsychologie) that, instead of philosophical, potentially makes “form” cognitive in a surprisingly contemporary way; and (3) to perhaps in the process rejuvenate theoretical interest in the status of “form” separate from its application.

Backhaus (1999) argued that Simmel’s formal sociology has an “affinity with” the phenomenology of Husserl, in particular the intentional relationality of mental acts, or the structures of pure consciousness (eides) that, in Simmel’s case, apply to forms of association. Instead of identifying empirical patterns or correlations, formal sociology registers the “cognition of an eidetic structure” (e.g. of “competition,” “conflict,” or “marriage”) (Backhaus 273). Like Husserl’s phenomenology, Simmel identifies these structures as transcendent in relation to particular, sensible and empirical instantiations; but he also does not suggest that forms are “empirical universals” that do not vary according to their instantiations or are not independent from them. If that were the case, then formal sociology would be an empirical science with a “body of collective positive content” that predetermines what can and cannot be present in a specific empirical setting and therefore what counts as having a “legitimate epistemic status” (such as the causes and effects of conflict). Simmel’s emphasis, by contrast, focuses on the analysis of form as it exhibits a “necessary structure” and allows the empirical “given” to appear as it does (Backhaus 264).

More generally, Backhaus concludes as follows:

The attempt to fit Simmel’s a priori structures of the forms of association into a Kantian formal a priori is not possible. Both … interactional and cognitive structures characterize the objects of sociological observation and are not structures inherent to the subjective conditions of the observer (Backhaus 262).

Backhaus’ argument here has given a certain license to formal sociology to spread beyond the friendly confines of network analysis. That spread is contingent on finding forms “not constituted by transactions but instead [giving] form to transactions—because they posit discrete, pregiven, and fixed entities that exist outside of the material plane prior to their instantiation” (Erikson 2013: 225). To posit these entities does not require finding a cognitive structure for the purposes of meaningful synthesis (in Kantian pure cognition). Simmel refers to forms of sociation as instead “[residing] a priori in the elements themselves, through which they combine, in reality, into the synthesis, society” (1971: 38).

So here is the puzzle. If we follow Backhaus’ lead and not read forms of sociation as Kantian categories, then we commit (eo ipso) to a priori elements as part of social relations, not simply in faculties of reason. How is that possible? Backhaus interprets this as being equivalent to the material a priori proposed by Husserl, in which forms of sociation are analogous to intentional objects (1999: 262). In principle, there is much to recommend this argument, not least that it resonates with Simmel’s methodological pluralism vis-a-vis form (Levine 1998). However, the best that Backhaus can do to support a Husserl/Simmel connection is to say that Simmel’s thought has an “affinity with” Husserl’s phenomenology. As he writes elsewhere:

Simmel was neither collaborator nor student of Husserl, and Simmel’s works appear earlier than the Husserlian influenced philosophers who were to become the first generation phenomenologists. Based on the supposition that Simmel’s later thought does parallel Husserl’s, can it be said that Simmel was coming to some of the same conclusions as Husserl, but yet did not recognize that what he was doing was unfolding an emergent philosophical orientation? An affirmative answer appears plausible. Yet, it is likely that Husserl was an influence on Simmel, without receiving public acknowledgement, since Simmel infrequently cites other thinkers within the body of his texts or within his limited use of footnotes (2003: 223-224).

And yet there is no available evidence (to date) that can document a direct influence of Husserl’s phenomenology on Simmel’s theory of forms (and/or vice versa). Beyond this, the timelines for such an influence do not exactly match, although Simmel and Husserl were contemporaries and, by all accounts, friends. While they did exchange letters, of the ones that survive there is (at least according to one interpretation) nothing of “philosophical value” in them (Staiti 2004: 173; though see Goodstein 2017: 18n9). Simmel’s concern with “psychology” long predates the publication of Husserl’s Logical Investigations in 1900-01. Simmel’s Philosophy of Money was published around the same time (1900) and marked his most extensive engagement with formal sociology by that point (as Simmel called it, “the first work … that is really my work”). Husserl, however, does not discuss the material a priori in Logical Investigations. In fact, the key source for Husserl’s claims about it doesn’t appear until much later: his 1919-1927 Natur und Geist lectures (Staiti 2004: chap 5). While Husserl does discuss “eidetic ontologies” in the first volume of Ideas Pertaining to a Pure Phenomenology and to a Phenomenological Philosophy (1912), written during Husserl’s tenure at the University of Gottingen (1901-1916), it seems relevant that Simmel’s two key discussions of “form” (in the Levine reader: “How is Society Possible?” and “The Problem of Sociology”) are both found in Simmel’s 1908 Soziologie: Untersuchungen uber die Formen der Vergesellschaftung and draw from material that appears much earlier (Goodstein 2017: 66).

None of this omits or definitively puts to rest an influence of Husserl on Simmel that, as Backhaus suggests, goes uncited and cannot be traced through published work. All of these details of a connection still without an authoritative answer makes Goodstein (2017: 18n9) propose tracking down the personal and intellectual relationship between Husserl and Simmel as a “good dissertation topic.” At the very least, this suggests that there might be more to the story apropos the status of “form” than we understand at this point and which is enough to reopen a (seemingly) closed case on which formal sociology (at least partially) rests, making this about a lot more than just an obscure footnote in the boring annals of sociology. It also seems relevant to emphasize a possible different reason why Husserl’s eidos and Simmel’s forms seem so similar, but in fact are not.

There is a definite parallel between Husserl and Simmel in that they both took positions against experimental psychology at the time. However, to assume that this means they both took the same position (which, in this case, would be one that Husserl would be credited with making, and which was against “psychologism” in toto) could make the most sense in retrospect only because the historical context has not yet been thoroughly described enough to allow us to see a different position available at the time, one whose content could be described (in the negative) as not experimental psychology, not phenomenology and not descriptive psychology. On these terms, this remains effectively a non-position in the present-day disciplinary landscape, with experimental psychology, phenomenology and descriptive psychology (qua culture) all being more or less still recognizable between now and then. This is only true, however, if we omit a nascent position (still) to be made now, possibly as cognitive social science (see Lizardo 2014), and which was available then as volkerpsychologie.

All of this suggests contextual reasons not to settle for reading Simmel as a phenomenologist. What I want to propose is that there are also further biographical reasons only recently come to light. Elizabeth Goodstein points in the direction with her insight that when Simmel uses the term “‘a priori … this usage … extends the notion of epistemological prerequisites to include their cultural-psychological and sociological formation [which] had its intellectual roots in Volkerpsychologie” (Goodstein 2017: 65; see also Frisby 1984). Goodstein here draws from the late German scholar Klaus Kohnke in what is arguably the most authoritative source on Simmel’s early influences: Kohnke’s untranslated Der junge Simmel in Theoriebeziehungen und sozialen Bewegungen (1996). Goodstein interprets this reading of Simmel’s a priori (both non-Kantian and non-phenomenological) as “[recognizing] the constructive role of culture and narrative framework in constituting and maintaining knowledge practices” (65). Even this is not completely satisfactory, however, as Kohnke (1990) himself suggests by observing the direct influence of volkerpsychologie on Simmel’s appropriation of two of its major themes—“condensation” and “apperception”—which can be categorized as “cultural” (in any contemporary meaning of the word) only very partially (see also Frisby 1984).

So where are we? Simmel’s a priori is essential to formal sociology, but it is not Kantian. We also have little reason to believe that is phenomenological, though this currently provides its best defense. It also cannot be translated as cultural, at least not in a contemporary sense. What we are left with is the influence of volkerpsychologie as part of Simmel’s intellectual history.

We are helped in defining volkerpsychologie by the fact that it has recently become a topic of conversation among historians of science (see Hopos Spring 2020). This interest has been piqued by a recognition of volkerpsychologie as a kind of interscientific space in the developing universe of the human sciences in the 19th century. Specifically, it was not experimental psychology (Wilhelm Wundt) and not descriptive psychology (Wilhelm Dilthey). In the latter sense, it was not an antidote to experimentalism and did not center around “understanding.” In the former sense, it promoted an explanatory framework but outside of the laboratory. Officially, Volkerpsychologie was initiated by the philosophers and philologists Moritz Lazarus and Heymann Steinthal in the mid-19th century. When Simmel entered Berlin University in 1876, his initial interest was history, studying with Theodor Mommsen. His interests soon shifted to psychology, however, and Lazarus became his main teacher.

The subsequent influence of Lazarus and Steinthal on Simmel is clear. Much of Simmel’s initial work in the early 1880s (including his rejected dissertation on music; Simmel [1882] 1968) was published in the journal that Lazarus and Steinthal founded and edited: Zeitschrift für Völkerpsychologie und Sprachwissenschaft (Reiners 2020). Simmel sent his essay (1892) on a nascent sociologie (“Das probleme der sociologie”) to Lazarus on his seventieth birthday, adding a letter in which Simmel wrote that the essay constituted “the most recent result of lines of thought that you first awakened in me. For however, divergent my subsequent development became, I shall nonetheless never forget that before all others, you directed me to the problem of the superindividual and its depths, whose investigation will probably fill out the productive time that remains to me” (quoted in Goodstein 2017: 65). In 1891, Steinthal directed readers of the journal that replaced Volkerpsychologie (Zeitschrift des Vereins für Volkskunde) to the “work of Georg Simmel” in order to see how volkerpsychologie and the nascent field of sociology both search for “the psychological processes of human society” (Kusch 2019: 264).

If Simmel was influenced by volkerpsychologie, he was far from alone (Klautke 2013). Durkheim was familiar with the volkerpsychologie, particularly the work of Lazarus and Steinthal. In fact, he cites (1995/1912:12n14) volkerpsychologie in the Elementary Forms as “putting the hypothesis first for “mental constitution [as depending] at least in part upon historical, hence social factors … Once this hypothesis is accepted, the problem of knowledge can be framed in new terms.” Durkheim references the Zeitschrift für Völkerpsychologie and mentions Steinthal in particular. Franz Boas (1904), meanwhile, gives “special mention” to volkerpsychologie as being a major influence on the history of anthropology for proposing “psychic actions that take place in the individual as a social unit,” also referencing the work of Steinthal (520). For his part, Bronislaw Malinowski had studied with Wundt in Leipzig and started an (unfinished) dissertation in volkerpsychologie (Forster 2010: 204ff). Boas and Malinowski provide a direct link from Lazarus and Steinhal’s volkerpsychologie to the “culture concept” (see Stocking 1966; Kalmar 1987). Mikhail Bakhtin also mentions Lazarus and Steinhal’s volkerpsychologie as an influence on his definition of dialogics and speech genres or “problems of types of speech.” Volkerpsychologie anticipates “a comparable way of conceptualizing collective consciousness” (see Reitan 2010).

This historiography thus finds the influence of volkerpsychologie on a variety of recognized disciplines and influences that reach into the present. More recent efforts are able to distinguish that influence from the influence of descriptive psychology, which is well-documented. Volkerpsychologie constituted a space of possibility in human science that did not settle into the disciplinary arrangement of the research university that still persists largely unchanged into the present (Clark 2008). As Goodstein (2017) notes, Simmel himself mirrors this with an oeuvre that remains unrecognizable from any single disciplinary guise. If Simmel did not identify with volkerpsychologie when certain bureaucratic requirements required him to declare a scholarly identity, this was at least partially because of the association of volkerpsychologie with scholars of Jewish heritage (including Lazarus and Steinthal), combined with prevailing anti-Semitism, with which Simmel was all too familiar (Kusch 2019: 267ff). Volkerpsychologie itself would later be terminologically appropriated by the Nazified “volk” which further contributed to the erasure of its 19th century history.

The purpose of recounting this history (obscure no doubt) is to perhaps rejuvenate interest in Simmel’s formal approach as more appropriately situated within a disciplinary space that anticipates cognitive social science. The ramifications of this are far beyond the scope of this post to draw out in sufficient detail. That will be saved for a later post (maybe). To close, I’ll just sketch one possible implication, using Omar’s recent distinction between “cognitive” and “cultural kinds.”

To make that distinction requires some way of distinguishing the cognitive from the cultural, i.e. giving it a “mark.” The philosopher Mark Rowlands (2013: 212) attempts this as follows: what marks the cognitive is “(1) the manipulation and transformation of information-bearing structures, where this (2) has the proper function of making available, either to the subject or to subsequent processing operations, information that was hitherto unavailable, where (3) this making available is achieved by way of the production, in the subject of the process, of a representational state and (4) the process belongs to a cognitive subject.” Rowlands subscribes to extended, enactive, embodied and embedded (4Es) cognition in making this argument, in which the key claim is not about “the mind” but about “mental phenomena.”

The proposal here is that a volkerpsychologie reading could be more accurate in situating “form” as having something more like a “mark of the cognitive” than the material a priori. For his part, Backhaus (1999) is careful to bracket the level of eidos from what he calls psychological associations and empirical universals. Perhaps, what would be identified as form could be empirically identified as carrying a cognitive content as “information-bearing structures.” This suggests an alternate way of finding a priori conditions in social relations. The problem is that this would commit a far more egregious “reading into” Simmel than reading Husserl into him. Any such effort would erase the historicism that guides my critique of Backhaus.

However, to the degree that volkerpsychologie is situated in a similar disciplinary space as cognitive social science (akin to 4Es cognition) this might lessen the violation. One historical effort (Kusch 2019) reads much of the original German-language research, published alongside Simmel’s own, and finds general commitments to relativism and materialism, meaning that (following the “strong” version of Lazarus and Steinthal) volkerpsychologie finds apperceptions “compressed” in even unproblematic forms of consciousness and locates these in an “objective spirit” as language, institutions and tools. Stronger versions also took umbrage with a normative application of volkerpsychologie because this arbitrarily bracketed an explanatory focus that endorsed only a relativist metaphysics (to an empirical context). Stronger versions even took a de facto Kantian critique a step further in attempting psychological explanations for what could be posited through logical inference (like freedom of the will). This did not mean resorting to cultural explanation, however. In fact, Dilthey distanced himself from volkerpsychologie because of its explanatory thrust. He developed his more “descriptive” approach (in part) in opposition to this. Strong versions of volkerpsychologie attempted generative explanations of intuitions derived from an original (empirical) context.

If there is any legitimate parallel between volkerpsychologie and formal sociology, then “form” could be given an entirely different treatment: conveying cognitive kinds that, among other things, allow for instances of particular cultural kinds.

References

Backhaus, Gary. (1999). “Georg Simmel as an Eidetic Social Scientist.” Sociological Theory 16: 260-281.

____. (2003). “Husserlian Affinities in Simmel’s Later Philosophy of History: The 1918 Essay.” Human Studies 26: 223-258.

Clark, William. (2008). Academic Charisma and the Origins of the Modern Research University. UChicago Press.

Erikson, Emily. (2013). “Formalist and Relationalist Theory in Social Network Analysis.” Sociological Theory 31: 219-242.

Frisby, David. (1984). “Georg Simmel and Social Psychology.” History of the Behavioral Sciences 20: 107-127.

Goodstein, Elizabeth. (2017). Georg Simmel and the Disciplinary Imagination. Stanford UP.

Hopos (Special Issue: Descriptive Psychology and Volkerpsychologie: In the Contexts of Historicism, Relativism and Naturalism). Spring 2020.

Jaworski, Gary. (1990). “Robert Merton’s Extension of Simmel’s Ubersehbar.” Sociological Theory 8: 99-105.

Kalmar, Ivan. (1987). The Völkerpsychologie of Lazarus and Steinthal and the Modern Concept of Culture. Journal of the History of Ideas 48: 671-690.

Klautke, Egbert. (2013). “The French reception of Völkerpsychologie and the origins of the social sciences.” Modern Intellectual History 10: 293-316.

Kohnke, Klaus. (1990). “Four Concepts of Social Science at Berlin University: Dilthey, Lazarus, Schmoller and Simmel.” in Georg Simmel and Contemporary Sociology.

Kusch, Martin (2019). “From Volkerpsychologie to the Sociology of Knowledge.” Hopos 9: 250-274.

Lizardo, Omar. (2014). “Beyond the Comtean Schema: The Sociology of Culture and Cognition Versus Cognitive Social Science.” Sociological Forum 29: 983-989.

Reiners, Stefan. (2020). “Our Science Must Establish Itself”: On the Scientific Status of Lazarus and Steinthal’s Völkerpsychologie.” Hopos 10: 234-253.

Rowlands, Mark. (2013). The New Science of the Mind: From Extended Mind to Embodied Phenomenology. MIT Press.

Silver, Daniel and Monica Lee. (2012). “Self-relations in Social Relations.” Sociological Theory 30: 207-237.

Silver, Daniel and Milos Brocic. (2019). “Three Concepts of Form in Simmel’s Sociology.” The Germanic Review 94: 114-124.

Stocking, George. (1966). “Franz Boas and the Culture Concept in Historical Perspective” American Anthropologist 68: 867-882.

Staiti, Andrea (2004). Husserl’s Transcendental Phenomenology: Nature, Sprit and Life. Cambridge U Press.