The Eliminativistic Implicit (I): The Necker Cube of Everyday and Scientific Explanation

by rsbakker

Go back to what seems the most important bit, then ask the Intentionalist this question: What makes you think you have conscious access to the information you need? They’ll twist and turn, attempt to reverse the charges, but if you hold them to this question, it should be a show-stopper.

What follows, I fear, is far longer winded.

Intentionalists, I’ve found, generally advert to one of two general strategies when dismissing eliminativism. The first is founded on what might be called the ‘Preposterous Complaint,’ the idea that eliminativism simply contradicts too many assumptions and intuitions to be considered plausible. As Uriah Kriegal puts it, “if eliminativism cannot be acceptable unless a relatively radical interpretation of cognitive science is adopted, then eliminativism is not in good shape” (“Non-phenomenal Intentionality,” 18). But where this criticism would be damning in other, more established sciences, it amounts to little more than an argument ad populum in the case of cognitive science, which as of yet lacks any consensual definition of its domain. The very naturalistic inscrutability behind the perpetual controversy also motivates the Eliminativist’s radical interpretation. The idea that something very basic is wrong with our approach to questions of experience and intentionality is by no means a ‘preposterous’ one. You could say the reality and nature of intentionality is the question. The Preposterous Complaint, in other words, doesn’t so much impugn the position as insinuate career suicide.

The second turns on what might be called the ‘Presupposition Complaint,’ the idea that eliminativism implicitly presupposes the very intentionality that it claims to undermine. The tactic generally consists of scanning the eliminativist’s claims, picking out various intentional concepts, then claiming that use of such concepts implicitly affirms the existence of intentionality. The Eliminativist, in other words, commits ‘cognitive suicide’ (as Lycan, 2005, calls it). Insofar as the use of intentional concepts is unavoidable, and insofar as the use of intentional concepts implicitly affirms the existence of intentionality, intentionality is ineliminable. The Eliminativist is thus caught in an obvious contradiction, explicitly asserting not-A on the hand, while implicitly asserting A on the other.

On BBT, intentionality as traditionally theorized, far from simply ‘making explicit’ what is ‘implicitly the case,’ is actually a kind of conceptual comedy of errors turning on heuristic misapplication and metacognitive neglect. Such appeals to ‘implicit intentionality,’ in other words, are appeals to the very thing BBT denies. They assume the sufficiency of the very metacognitive intuitions that positions such as my own call into question. The Intentionalist charge of performative contradiction simply begs the question. It amounts to nothing more than the bald assertion that intentionality cannot be eliminated because intentionality is ineliminable.

The ‘Presupposition Complaint’ is pretty clearly empty as an argumentative strategy. In dialogical terms, however, I think it remains the single biggest obstacle to the rational prosecution of the Intentionalist/Eliminativist debate—if only because of the way it allows so many theorists to summarily dismiss the threat of Eliminativism. Despite its circularity, the Presupposition Complaint remains the most persistent objection I encounter—in fact, many critics persist in making it even after its vicious circularity has been made clear. And this has led me to realize the almost spectacular importance of the notion of the implicit plays in all such debates. For many thinkers, the intentional nature of the implicit is simply self-evident, somehow obvious to intuition. This is certainly how it struck me before I began asking the kinds of questions motivating the present piece. After all, what else could the implicit be, if not the intentional ‘ground’ of our intentional ‘practices’?

In what follows, I hope to show how this characterization of the implicit, far from obvious, actually depends, not only on ignorance, but on a profound ignorance of our ignorance. On the account I want to give here, the implicit, far from naming some spooky ‘infraconceptual’ or ‘transcendental’ before of thought and cognition, simply refers to what we know is actually occluded from metacognitive appraisals of experience: namely, nature as described by science. To frame the issue in terms of a single question, what I want to ask in this post and its sequels is, What warrants the Intentionalist’s claims regarding implicit normativity, say, over an Eliminativist’s claims of implicit mechanicity?

So what is the implicit? Given the crucial role the concept plays in a variety of discourses, it’s actually remarkable how few theorists have bothered with the question of making the implicit qua implicit explicit (Stephen Turner and Eugene Gendlin are signature exceptions in this regard, of course). Etymologically, ‘implicit’ derives from the Latin, implicitus, the participle of implico, which means ‘to involve’ or ‘to entangle,’ meanings that seem to bear more on implicit’s perhaps equally mysterious relatives, ‘imply’ or ‘implicate.’ According to Wikitionary, uses that connote ‘entangled’ are now obsolete. Implicit, rather, is generally taken to mean, 1) “Implied directly, without being directly expressed,” 2) “Contained in the essential nature of something but not openly shown,” and 3) “Having no reservations or doubts; unquestioning or unconditional; usually said of faith or trust.” Implicit, in other words, is generally taken to mean unspoken, intrinsic, and unquestioned.

Prima facie, at least, these three senses are clearly related. Unless spoken about, the implicit cannot be questioned, and so must remain an intrinsic feature of our performances. The ‘implicit,’ in other words, refers to something operative within us that nonetheless remains hidden from our capacity to consciously report. Logical or material inferential implications, for instance, guide subsequent transitions within discourse, whether we are conscious of them or not. The same might be said of ‘emotional implications,’ or ‘political implications,’ or so on.

Let’s call this the Hidden Constraint Model of the implicit, the notion that something outside conscious experience somehow ‘contains’ organizing principles constraining conscious experience. The two central claims of the model can be recapitulated as:

1) The implicit lies in what conscious cognition neglects. The implicit is inscrutable.

2) The implicit somehow constrains conscious cognition. The implicit is effective.

From inscrutability and effectiveness, we can infer at least two additional features pertaining to the implicit:

3) The effective constraints on any given moment of conscious cognition require a subsequent moment of conscious cognition to be made explicit. We can only isolate the biases specific to a claim we make subsequent to that claim. The implicit, in other words, is only retrospectively accessible.

4) Effective constraints can only be consciously cognized indirectly via their effects on conscious experience. Referencing, say, the ‘implicit norms governing interpersonal conduct’ involves referencing something experienced only in effect. ‘Norms’ are not part of the catalogue of nature—at least as anything recognizable as such. The implicit, in other words, is only inferentially accessible.

So consider, as a test case, Hume’s famous meditations on causation and induction. In An Enquiry Concerning Human Understanding, Hume points out how reason, no matter how cunning, is powerless when it comes to matters of fact. Short of actual observation, we have no way of divining the causal connections between events. When we turn to experience, however, all we ever observe is the conjunction of events. So what brings about our assumptive sense of efficacy, our sense of causal power? Why should repeating the serial presentation of two phenomena produce the ‘feeling,’ as Hume terms it, that the first somehow determines the second? Hume’s ‘skeptical solution,’ of course, attributes the feeling to mere ‘custom or habit.’ As he writes, “[t]he appearance of a cause always conveys the mind, by a customary transition, the idea of an effect” (ECHU, 51, italics my own).

All four of the features enumerated above are clearly visible in the above. Hume makes no dispute of the fact that the repetition of successive events somehow produces the assumption of efficacy. “On this,” he writes, “are founded all our reasonings concerning matters of fact or existence” (51). Exposure to such repetitions fundamentally constrains our understanding of subsequent exposures, to the point where we cannot observe the one without assuming the other—to the point where the bulk of scientific knowledge is raised upon it. Efficacy is effective—to say the least!

But there’s nothing available to conscious cognition—nothing observable in these successive events—over and above their conjunction. “One event follows another,” Hume writes; “but we never can observe any tie between them. They seem conjoined, but never connected” (49). Efficacy, in other words, is inscrutable as well.

So then what explains our intuition of efficacy? The best we can do, it seems, is to pause and reflect upon the problem (as Hume does), to posit some X (as Hume does) reasoning from what information we can access. Efficacy, in other words, is only retrospectively and inferentially accessible.

We typically explain phenomena by plugging them into larger functional economies, by comprehending how their precursors constrain them and how they constrain their successors in turn. This, of course, is what made Hume’s discovery—that efficacy is inscrutable—so alarming. When it comes to environmental inquiries we can always assay more information via secondary investigation and instrumentation. As a result, we can generally solve for precursors in our environments. When it comes to metacognitive inquiries such as Hume’s, however, we very quickly stumble into our own incapacity. “And what stronger instance,” Hume asks, “can be produced of the surprising ignorance and weakness of the understanding, than the present?” (51). Efficacy, the very thing that binds phenomena to their precursors, is itself without precursors.

Not surprisingly, the comprehension of cognitive phenomena (such as efficacy) without apparent precursors poses a special kind of problem. Given efficacy, we can comprehend environmental nature. We simply revisit the phenomena and infer, over and over, accumulating the information we need to arbitrate between different posits. So how, then, are we supposed to comprehend efficacy? The empirical door is nailed shut. No matter how often we revisit and infer, we simply cannot accumulate the data we need to arbitrate between our various posits. Above, we see Hume rooting around with questions, (our primary tool for making ignorance visible) and finding no trace of what grounds his intuitions of empirical efficacy. Thus the apparent dilemma: Either we acknowledge that we simply cannot understand these intuitions, “that we have no idea of connexion or power at all, and that these words are absolutely without any meaning” (49), or we elaborate some kind of theoretical precursor, some fund of hidden constraint, that generates, at the very least, the semblance of knowledge. We posit some X that ‘reveals’ or ‘expresses’ or ‘makes explicit’ the hidden constraint at issue.

These ‘X posits’ have been the bread and butter of philosophy for some time now. Given Hume’s example it’s easy to see why: the structure and dynamics of cognition, unlike the structure and dynamics of our environment, do not allow for the accumulation of data. The myriad observational opportunities provided by environmental phenomena simply do not exist for phenomena like efficacy. Since individual (and therefore idiosyncratic) metacognitive intuitions are all we have to go on, our makings explicit are pretty much doomed to remain perpetually underdetermined—to be ‘merely philosophical.’

I take this as uncontroversial. What makes philosophy philosophy as opposed to a science is its perennial inability to arbitrate between incompatible theoretical claims. This perennial inability to arbitrate between incompatible theoretical claims, like the temporary inability to arbitrate between incompatible theoretical claims in the sciences, is in some important respect an artifact of insufficient information. But where the sciences generally possess the resources to accumulate the information required, philosophy does not. Aside from metacognition or ‘theoretical reflection,’ philosophy has precious little in the way of informational resources.

And yet we soldier on. The bulk of traditional philosophy relies on what might be called the Accessibility Conceit: the notion that, despite more than two thousand years of failure, retrospective (reflective, metacognitive) interrogations of our activities somehow access enough information pertaining to their ‘intrinsic character’ to make the inferential ‘expression’ of our implicit precursors a viable possibility. Hope, as they say, springs eternal. Rather than blame their discipline’s manifest institutional incapacity on some more basic metacognitive incapacity, philosophers generally blame the problem on the various conceptual apparatuses used. If they could only get their concepts right, the information is there for the taking. And so they tweak and they overturn, posit this precursor and that, and the parade of ‘makings explicit’ grows and grows and grows. In a very real sense, the Accessibility Conceit, the assumption that the tools and material required to cognize the implicit are available, is the core commitment of the traditional philosopher. Why show up for work, otherwise?

The question of comprehending conscious experience is the question of comprehending the constitutive and dynamic constraints on conscious experience. Since those constraints don’t appear within conscious experience, we pay certain people called ‘philosophers’ to advance speculative theories of their nature. We are a rather self-obsessed species, after all.

Advancing speculative hypotheses regarding each other’s implicit nature is something we do all the time. According to Robin Dunbar, some two thirds of human communication is devoted to gossip. We are continually replaying, revisiting—even our anticipations yoke the neural engines of memory. In fact, we continually interrogate our emotionally charged interactions, concocting rationales, searching for the springs of others’ actions, declaring things like ‘She’s just jealous,’ or ‘He’s on to you.’ There is, you might say, an ‘Everyday Implicit’ implicit in our everyday discourse.

As there has to be. Conscious experience may be ‘as wide as the sky,’ as Dickinson says, but it is little more than a peephole. Conscious experience, whatever it turns out to be, seems to be primarily adapted to deliberative behaviour in complex environments. Among other things, it operates as a training interface, where the deliberative repetition of actions can be committed to automatic systems. So perhaps it should come as no surprise that, like behaviour, it is largely serial. When peephole, serial access to a complex environment is all you have, the kind of retrospective inferential capacity possessed by humans becomes invaluable. Our ability to ‘make things explicit’ is pretty clearly a central evolutionary design feature of human consciousness.

In a fundamental sense, then, making-explicit is just what we humans do. It makes sense that with time, especially once literacy allowed for the compiling of questions—an inventory of ignorance, you might say—that we would find certain humans attempting to make making explicit itself explicit. And since making each other explicit was something that we seemed to do with some degree of reliability, it makes sense that the difficulty of this new task should confound these inquirers. The Everyday Implicit was something they used with instinctive ease, reliably attributing all manner of folk-intentional properties to individuals all the time. And yet, whenever anyone attempted to make this Everyday Implicit explicit, they seemed to come up with something different.

No one could agree on any canonical explication. And yet, aside from the ancient skeptics, they all agreed on the possibility of such a canonical explication. They all hewed to the Accessibility Conceit. And since the skeptics’ mysterian posit was as underdetermined as any of their own claims, they were inclined to be skeptical of the skeptics. Otherwise, their Philosophical Implicit remained the only game in town when it came to things human and implicit. They need only look to the theologians for confirmation of their legitimacy. At least they placed their premises before their conclusions!

But things have changed. Over the past few decades, cognitive scientists have developed a number of ingenious experimental paradigms designed to reveal the implicit underbelly of what we think and do. In the now notorious Implicit Association Test, for instance, the time subjects require to pair concepts is thought to indicate the cognitive resources required, and thus provide an indirect measure of implicit attitudes. If it takes a white individual longer to pair stereotypically black names with positive attributes than it does white names, this is presumed to evidence an ‘implicit bias’ against blacks. Actions, as the old proverb has it, speak louder than words. It does seem intuitive to suppose that the racially skewed effort involved in value identifications tokens some kind of bias. Versions of this of this paradigm continue to proliferate. Once the exclusive purview of philosophers, the implicit has now become the conceptual centerpiece of a vast empirical domain. Cognitive science has now revealed myriad processes of implicit learning, interpretation, evaluation, and even goal-setting. Taken together, these processes form what is generally referred to as System 1 cognition (see table below), an assemblage of specialized cognitive capacities—heuristics—adapted to the ‘quick and dirty’ solution of domain specific ‘problem ecologies’ (Chow, 2011; Todd and Gigerenzer, 2012), and which operate in stark contrast to what is called System 2 cognition, the slow, serial, and deliberate problem solving related to conscious access (defined in Dehaene’s operationalized sense of reportability)—what we take ourselves to be doing this very moment, in effect.


System 1 Cognition (Implicit) System 2 Cognition (Explicit)
Not conscious Conscious
Not human specific Human specific
Automatic Deliberative
Fast Slow
Parallel Sequential
Effortless Effortful
Intuitive Reflective
Domain specific Domain general
Pragmatic Logical
Associative Rulish
High capacity Low capacity
Evolutionarily old Evolutionarily young

* Adapted from Frankish and Evans, “The duality of mind: A historical perspective.”

What are called ‘dual process’ or ‘dual system’ theories of cognition are essentially experimentally driven complications of the crude dichotomy between unconscious/implicit and conscious/explicit problem solving that has been pondered since ancient times. As granular as this emerging empirical picture remains, it already poses a grave threat to our traditional explicitations of the implicit. Our cognitive capacities, it turns out, are far more fractionate, contingent, and opaque than we ever imagined. Decisions can be tracked prior to a subject’s ability to report them (Haynes, 2008; or here). The feeling of willing can be readily tricked, and thus stands revealed as interpretative (Wegner, 2002; Pronin, 2009). Memory turns out to be fractionate and nonveridical (See Bechtel, 2008, for review). Moral argumentation is self-promotional rather than truth-seeking (Haidt, 2012). Various attitudes appear to be introspectively inaccessible (See Carruthers, 2011, for extensive review). The feeling of certainty has a dubious connection to rational warrant (Burton, 2008). The list of such findings continually grows, revealing an ‘implicit’ that consistently undermines and contradicts our traditional and intuitive self-image—what Sellars famously termed our Manifest Image.

As Frankish and Evans (2009) write in their historical perspective on dual system theories:

“The idea that we have ‘two minds’ only one of which corresponds to personal, volitional cognition, has also wide implications beyond cognitive science. The fact that much of our thought and behaviour is controlled by automatic, subpersonal, and inaccessible cognitive processes challenges our most fundamental and cherished notions about personal and legal responsibility. This has major ramifications for social sciences such as economics, sociology, and social policy. As implied by some contemporary researchers … dual process theory also has enormous implications for educational theory and practice. As the theory becomes better understood and more widely disseminated, its implications for many aspects of society and academia will need to be thoroughly explored. In terms of its wider significance, the story of dual-process theorizing is just beginning.” 25

Given the rhetorical constraints imposed by their genre, this amounts to the strident claim that a genuine revolution in our understanding of the human is underway, one that could humble us out of existence. The simple question is, Where does that revolution end?

Consider what might be called the ‘Worst Case Scenario’ (WCS). What if it were the case that conscious experience and cognition have evolved in such a way that the higher dimensional, natural truth of the implicit utterly exceeds our capacity to effectively cognize conscious experience and cognition outside a narrow heuristic range? In other words, what if the philosophical Accessibility Conceit were almost entirely unwarranted, because metacognition, no matter how long it retrospects or how ingeniously it infers, only accesses information pertinent to a very narrow band of problem solving?

Now I have a number of arguments for why this is very likely the case, but in lieu of those arguments, it will serve to consider the eerie way our contemporary disarray regarding the implicit actually exemplifies WCS. People, of course, continue using the Everyday Implicit the way we always have. Philosophers continue positing their incompatible versions of the Philosophical Implicit the way they have for millennia. And scientists researching the Natural Implicit continue accumulating data, articulating a picture that seems to contradict more and more of our everyday and philosophical intuitions as it gains dimensionality.

Given WCS, we might expect the increasing dimensionality of our understanding would leave the functionality of the Everyday Implicit intact, that it would continue to do what it evolved to do, simply because it functions the way it does regardless of what we learn. At the same time, however, we might expect the growing fidelity of the Natural Implicit would slowly delegitimize our philosophical explications of that implicit, not only because those explications amount to little more than guesswork, but because of the fundamental incompatibility of intentional and the causal conceptual registers.

Precisely because the Everyday Implicit is so robustly functional, however, our ability to gerrymander experimental contexts around it should come as no surprise. And we should expect that those invested in the Accessibility Conceit would take the scientific operationalization of various intentional concepts as proof of 1) their objective existence, and 2) the fact that only more cognitive labour, conceptual, empirical, or both, is required.

If WCS were true, in other words, one might expect that cognitive sciences invested in the Everyday and Philosophical Implicit, like psychology, would find themselves inexorably gravitating about the Natural Implicit as its dimensionality increased. One might expect, in other words, that the Psychological Implicit would become a kind of decaying Necker Cube, an ‘unstable bi-stable concept,’ one that would alternately appear to correspond to the Everyday and Philosophical Implicit less and less, and to the Natural Implicit more and more.

Part Two considers this process in more detail.