Eliminativist Interventions: Gallagher on Neglect and Free Will

by rsbakker

In the previous post I tried to show how the shape of the free will debate can be explained in terms of the differences and incompatibilities between source-sensitive (causal) and source-insensitive (intentional) cognition. Rather than employ the overdetermined term, ‘free will,’ I considered the problem in terms of ‘choice-talk,’ the cognitive systems and language we typically employ when reporting behaviour. I then try to show how this simple step sideways allows us to see the free will debate as a paradigmatic, intellectual ‘crash space,’ a point where the combination of heuristic neglect and intellectual innovation generates systematic cognitive illusions.

As it happened, I read Shaun Gallagher’s excellent Enactivist Interventions: Rethinking the Mind while picking at this piece, and lo, discovered that he too tackles the problem of free will. I wrote what follows as an inescapable consequence.

Gallagher’s approach to the question of free will is diagnostic, much like my own. First, he wants to characterize how the problem is canonically posed, the ‘common understanding of the question,’ then he wants to show how this characterization gets the problem wrong. Discussing Libet’s now infamous ‘free won’t’ experiment, he points out that “In the experimental situation we are asked to pay attention to all of the processes that we normally do not attend to, and to move our body in a way that we do not usually move it…” which is to say, precisely those processes choice-talk systematically neglects. As he writes:

“These experiments, however, and more generally the broader discussions of motor control, have nothing to tell us about free will per se. If they contribute to a justification of perceptual or epiphenomenal theories of how we control our movement, these are not theories that address the question of free will. The question of free will is a different question.”

But why is the free will question a different question? Gallagher offers two different reasons why the question of free will simply has no place in the neurophysiology of decision-making:

“The attempt to frame the question of free will in terms of these subpersonal processes—either to dismiss it or to save it—is misguided for at least two reasons. First, free will cannot be squeezed into the elementary timescale of 150–350 milliseconds; free will is a longer-term phenomenon and, I will argue, it involves consciousness. Second, the notion of free will does not apply primarily to abstract motor processes or even to bodily movements that make up intentional actions—rather it applies to intentional actions themselves, described at the most appropriate pragmatic level of description.”

Essentially, Gallagher’s offering his own level-of-description argument. The first reason choice-talk has no place in neurophysiological considerations is that it only applies to the time-scale of personal action, and not to the time-scale of neurophysiological processes. This seems a safe enough assumption, given the affinity of choice-talk with personal action more generally. The problem is that we already know that free will has no application in neurophysiology—that it is expunged. The question, rather, is whether source-talk applies to personal time-scales. And the problem, as we saw above, is that it most certainly does. We can scale up the consequences of Libet’s experiment, talk of the brain deciding before conscious awareness of our decisions. In fact, we do this whenever we use biomedical facts to assess responsibility. Certainly, we don’t want to go back to the days of condemning the character of kids suffering ADHD and the like.

Gallagher’s second reason is that choice-talk only applies to the domain of intentional actions. He introduces the activity of lizard hunting to give an example of the applicability and inapplicability of choice-talk (and hence, free will). What’s so interesting here, from a heuristic neglect standpoint, is the way his thinking continually skirts around the issue of source-insensitivity.

“I am not at all thinking about how to move my body—I’m thinking about catching the lizard. My decision to catch the lizard is the result of a consciousness that is embedded or situated in the particular context defined by the present circumstance of encountering the lizard, and the fact that I have a lizard collection. This is an embedded or situated reflection, neither introspective nor focused on my body. It is ‘a first-person reflective consciousness that is embedded in a pragmatically or socially contextualized situation.”

Gallagher’s entirely right that we systematically neglect our physiology in the course of hunting lizards. Choice-talk belongs to a source-insensitive regime of problem solving—Gallagher himself recognizes as much. We neglect the proximal sources of behaviour and experience, focusing rather on the targets of those sources. Because this regime exhibits source-insensitivity, it relies on select correlations, cues, to the systems requiring solution. A face, for instance, is a kind of cuing organ, allowing others to draw dramatic conclusions of the basis of the most skeletal information (think happy faces, or any ‘emoticon’). The physiology of the expression-animating brain completely eludes us, and yet we can make striking predictions regarding what it will do next given things like ancestral biological integrity and similar training. A happy face on a robot, on the other hand, could mean anything. This ecological dependence is precisely why source-insensitive cognitive tools are so situational, requiring the right cues in the right circumstances to reliably solve select sets of problems—or problem ecologies.

So, Gallagher is right to insist that choice-talk, which is adapted to solve in source-insensitive or ‘shallow’ cognitive ecologies, has no application in source-sensitive or ‘deep’ cognitive ecologies. After all, we evolved these source-insensitive modes because, ancestrally speaking, biological complexity made source-sensitive cognition of living systems impossible. This is why our prescientific ancestors could go lizard hunting too.

Gallagher is also largely right to say that sourcing lizard-hunting a la neuroscience has nothing to do with our experience of hunting lizards—so long as everything functions as it should. Sun-stroke is but one of countless, potential ‘choice-talk’ breakers here.

But, once again, the question is whether source-talk applies to the nature of lizard hunting—which it certainly does. How could it not? Lizard hunting is something humans do—which is to say, biological through and through. Biology causes us to see lizards. Biology also causes us (in astronomically complicated, stochastic ways) to hunt them.

Gallagher’s whole argument hinges on an apple and orange strategy, the insistence that placing neurophysiological apples in the same bushel as voluntary oranges fundamentally mistakes the segregate nature of oranges. On my account both choice-talk and source-talk possess their respective problem-ecologies while belonging to the same high-dimensional nature. Choice-talk belongs to a system adapted to source-insensitive solutions, and as such, possesses a narrow scope of application. Source-talk, on the other hand, possesses a far, far broader scope of application, so much so that it allows us to report the nature of choice-talk. This is what Libet is doing. His findings crash choice-talk because choice-talk actually requires source-neglect to function happily.

On Gallagher’s account, free will and neurophysiology occupy distinct ‘levels of description,’ the one belonging to ‘intentional action,’ and the other to ‘natural science.’ As with the problem ecology of choice-talk, the former level is characterized by systematic source-neglect. But where this systematic neglect simply demarcates the problem-ecology of choice-talk from that of source-talk in my account, in Gallagher it demarcates an ontologically exceptional, low-dimensional ecology, that of ‘first-person reflective consciousness… embedded in a pragmatically or socially contextualized situation.’

This where post-cognitivists, having embraced high-dimensional ecology, toss us back into the intractable lap of philosophy. Gallagher, of course, thinks that some exceptional twist of nature forces this upon cognitive science, one that the systematic neglect of sources in things like lizard hunting evidences. But once you acknowledge neglect, the way Gallagher does, you have no choice but to consider the consequences of neglect. Magicians, for instance, are masters at manipulating our intuitions via neglect. Suppress the right kind of information, and humans intuit exceptional entities and events. Is it simply a coincidence that we both suffer source-neglect and we intuit exceptional entities and events when reflecting on our behaviour?

How, for instance, could reflection hope to distinguish the inability to source from the absence of sources? Gallagher agrees that metacognition is ecological—that there is no such thing as the ‘disembodied intellect.’ “Even in cases where we are able to step back,” Gallagher writes, “to detach ourselves from the demands of the immediate environment, and to engage in a second-order, conceptual deliberation, this stepping back does not make thinking any less of an embodied/ intersubjective skill.” Stepping back does not mean stepping out, despite seeming that way. Human metacognition is radically heuristic, source-insensitive through and through. Deliberative reflection on the nature of experience cannot but systematically neglect sources. This is why we hallucinate ‘disembodied intellects’ in the first place! We simply cannot, given our radically blinkered metacognitive vantage, distinguish confounds pertaining to neglect from properties belonging to experience. (The intuition, in fact, cuts the other way, which is why the ball of discursive yarn needs to be unraveled in the first place, why post-cognitivism is post.)

Even though Gallagher relies on neglect to relativize choice-talk to a particular problem-solving domain (his ‘level of description’), he fails to consider the systematic role played by source-insensitivity in our attempts to cognize cognition. He fails, in other words, to consider his own theoretical practice in exhaustively ecological terms. He acknowledges that it has to be ecological, but fails to consider what this means. As a result, he trips into phenomenological and pragmatic versions of the same confounds he critiques in cognitivism. Disembodied intellects become disembodied embodied intellects.

To be embodied is to be high-dimensional, to possess nearly inexhaustible amounts natural information. To be embodied, in other words, is to be susceptible to source-sensitive cognition. Except, Gallagher would have you believe, when its not, when the embodiment involves intentionality, in which case, we are told, source-talk no longer applies, stranding us with the low-dimensional resources of source-insensitive cognition (which is to say, perpetual disputation). ‘Disembodied intellects’ (one per theorist) are traded for irreducible phenomenologies (one per theorist) and/or autonomous normativities (one per theorist), a whole new set of explananda possessing natures that, we are assured, only intentional cognition can hope to solve.

Gallagher insists that intentional phenomena are embodied, ‘implicit,’ as he likes to say, in this or that high-dimensional ecological feature, only at a ‘level of description’ that only intentional cognition can solve. The obvious problem, of course, is that the descriptive pairing of low-dimensional intentional phenomena like ‘free will’ with high-dimensional ecologies amounts to no more than a rhetorical device short some high-dimensional account of intentionality. Terms such as ‘implicit,’ like ‘emergent’ or ‘autopoietic,’ raise far more questions than they answer. How is intentionality ‘implicit’ in x? How does intentionality ’emerge’ from x? Short some genuine naturalization of intentionality, very little evidences the difference between Gallagher’s ‘embodiment’ and haunting—‘daimonic possession.’

The discursively fatal problem, however, is that intentional cognition, as source-insensitive, relies on strategic correlations to those natures—and thus has no application to the question of natures. These are ‘quick and dirty’ systems adapted to the economical solution of practical problems on the fly. Only neglect makes it seem otherwise. This is why post-cognitivism, like cognitivism more generally, cannot so much as formulate, let alone explain, its explananda in any consensus-commanding way. On Gallagher’s account, institutional philosophy remains firmly in charge of cognitive scientific theorization, and will continue to do so in perpetuity as a ‘philosophy of nature’ (and in this respect, he’s more forthright than Hutto and Myin, who rhetorically dress their post-cognitive turn as an ‘escape’ from philosophy).

Ecological eliminativism suffers neither of these problems. Choice-talk has its problem-ecology. Source-talk has its problem-ecology. The two evolved on separate tracks, but now, thanks to radical changes in human cognitive ecology, they find themselves cheek and jowl, causing the former to crash with greater and greater frequency. This crash occurs, not because people are confusing ‘ontologically distinct levels of description,’ one exceptional, the other mundane, but because the kind of source-neglect required by the former does not obtain the way it did ancestrally. We should expect, moreover, the frequency of these crashes to radically increase as cognitive science and its technologies continue to mature. Continued insistence on ontologically and/or functionally exceptional ‘levels of description’ all but blinds us to this looming crisis.

Having acknowledged the fractionate and heuristic nature of deliberative metacognition, having acknowledged source-neglect, Gallagher now needs to explain what makes his exceptionalism exceptional, why the intentional events and entities he describes cannot be explained away as artifacts of inevitable heuristic misapplication. He finds neglect useful, but only because he neglects to provide a fulsome account of its metacognitive consequences. It possesses a second, far sharper edge.

 

 

If Free-Will were a Heuristic…

by rsbakker

Ecological eliminativism provides, I think, an elegant way to understand the free-will debate as a socio-cognitive ‘crash space,’ a circumstance where ecological variance causes the systematic breakdown of some heuristic cognitive system. What follows is a diagnostic account, and as such will seem to beg the question to pretty much everyone it diagnoses. The challenge it sets, however, is abductive. In matters this abstruse, it will be the power to explain and synthesize that will carry the theoretical morning if not the empirical day.

As hairy as it is, the free-will debate, at least in its academic incarnation, has a trinary structure: you have libertarians arguing the reality of how decision feels, you have compatibilists arguing endless ways of resolving otherwise manifest conceptual and intuitive incompatibilities, and you have determinists arguing the illusory nature of how decision feels.

All three legs of this triumvirate can be explained, I think, given an understanding of heuristics and the kinds of neglect that fall out of them. Why does the feeling of free will feel so convincing? Why are the conceptualities of causality and choice incompatible? Why do our attempts to overcome this incompatibility devolve into endless disputation?

In other words, why is there a free-will debate at all? As of 10:33 AM December 17^th, 2019, Googling “free will debate” returned 575,000,000 hits. Looking at the landscape of human cognition, the problem of free will looms large, a place where our intuitions, despite functioning so well in countless other contexts, systematically frustrate any chance of consensus.

This is itself scientifically significant. So far as pathology is the royal road to function, we should expect that spectacular breakdowns such as these will hold deep lessons regarding the nature of human cognition.

As indeed they do.

So, let’s begin with a simple question: If free-will were a heuristic, a tool humans used to solve otherwise intractable problems, what would it’s breakdown look like?

But let’s take a step back for a second, and bite a very important, naturalistic bullet. Rather than consider ‘free-will’ as a heuristic, let’s consider something less overdetermined: ‘choice-talk.’ Choice-talk constitutes one of at least two ways for us humans to report selections between behaviours. The second, ‘source-talk,’ we generally use to report the cognition of high-dimensional (natural) precursors, whereas we generally use choice-talk to report cognition absent high-dimensional precursors.

As a cognitive mechanism, choice-talk is heuristic insofar as it turns a liability into an asset, allowing us to solve social problems low-dimensionally—which is to say, on the cheap. That liability is source insensitivity, our congenital neglect of our biological/ecological precursors. Human cognition is fundamentally structured by what might be called the ‘biocomplexity barrier,’ the brute fact that biology is too complicated to cognize itself high-dimensionally. The choice-talk toolset manages astronomically complicated biological systems—ourselves and other people—via an interactional system reliably correlated to the high-dimensional fact of those systems given certain socio-cognitive contexts. Choice-talk works given the cognitive ecological conditions required to maintain the felicitous correlation between the cues consumed and the systems linked to them. Undo that correlation and choice-talk, like any other heuristic mechanism, begins to break down.

Ancestrally, we had no means of discriminating our own cognitive constitution. The division of cognitive labour between source-sensitive and source-insensitive cognition is one that humans constitutively neglect: we have to be trained to discriminate it. Absent such discrimination, the efficacy of our applications turn on the continuity of our cognitive ecologies. Given biocomplexity, the application of source-sensitive cognition to intractable systems—and biological systems in particular—is not something evolution could have foreseen. Why should we possess the capacity to intuitively reconcile the joint application of two cognitive systems that, as far as evolution was concerned, would never meet?

As a source-insensitivity workaround, a way to cognize behaviour absent the ability to source that behaviour, we should expect choice-talk cognition to misfire when applied to behaviour that can be sourced. We should expect that discovering the natural causes of decisions will scuttle the intuition that those decisions were freely chosen. The manifest incompatibility between high-dimensional source-talk and low-dimensional choice-talk arises because the latter has been biologically filtered to function in contexts precluding the former. Intrusions of source-talk applicability, when someone suffers a head injury, say, could usefully trump choice-talk applicability.

Choice-talk, in fact, possesses numerous useful limits, circumstances where we suspend its application to better solve social problems via other tools. As radically heuristic, choice-talk requires a vast amount of environmental stage-setting in order to function felicitously, an ecological ‘sweet spot’ that’s bound to be interrupted by any number of environmental contingencies. Some capacity to suspend its application was required. Intuitively, then, source-talk trumps choice-talk when applied to the same behaviour. Since the biocomplexity barrier assured that each mode would be cued the way it had always been cued since time immemorial, we could, ancestrally speaking, ignore our ignorance and generally trust our intuitions.

The problem is that source-talk is omni-applicable. With the rise of science, we realized that everything biological can be high-dimensionally sourced. We discovered that the once-useful incompatibility between source-talk and choice-talk can be scotched with a single question: If everything can be sourced, and if sources negate choices, then how could we be free? Incompatibility that was once-useful now powerfully suggests choice-talk has no genuinely cognitive applicability anywhere. If choice-talk were heuristic, in other words, you might expect the argument that ‘choices’ are illusions.

The dialectical problem, however, is that human deliberative metacognition, reflection, also suffers source-insensitivity and so also consists of low-dimensional heuristics. Deliberative metacognition, the same as choice-talk, systematically neglects the machinery of decision making: reflection consistently reports choices absent sources as a result. Lacking sensitivity to the fact of insensitivity, reflection also reports the sufficiency of this reporting. No machinery is required. The absence of proximal, high-dimensional sources is taken for something real, ontologized, becoming a property belonging to choices. Given metacognitive neglect, in other words, reflection reports choice-talk as expressing some kind of separate, low dimensional ontological order.

Given this blinkered report, everything depends on how one interprets that ontology and its relation to the high-dimensional order. Creativity is required to somehow rationalize these confounds, which, qua confounds, offer nothing decisive to adjudicate between rationalizations. If choice-talk were a heuristic, one could see individuals arguing, not simply that choices are ‘real,’ but the kind of reality they possess. Some would argue that choice possesses a reality distinct from biological reality, that choices are somehow made outside causal closure. Others would argue that choices belong to biological reality, but in a special way that explains their peculiarity.

If choice-talk were heuristic, in other words, you would expect that it would crash given the application of source-cognition to behaviours it attempts to explain. You would expect this crash to generate the intuition that choice-talk is an illusion (determinism). You would expect attempts to rescue choice would either take the form of insisting on its independent reality (libertarianism), or its secondary reality (compatibilism).

Two heuristic confounds are at work, the first a product of the naïve application of source-talk to human decision-making, cuing us to report the inapplicability of choice-talk tout court, the second the product of the naïve application of deliberative metacognition to human decision-making, cuing us to report the substantive and/or functional reality of ‘choice.’

If choice-talk were heuristic, in other words, you would expect something that closely resembles the contemporary free-will debate. You could even imagine philosophers cooking up cases to test, even spoof, the ways in which choice-talk and source-talk are cued. Since choices involve options, for instance, what happens when we apply source-talk to only one option, leaving the others to neglect?

If choice-talk were heuristic, in other words, you could imagine philosophers coming up things like ‘Frankfurt-style counterexamples.’ Say I want to buy a pet, but I can’t make up my mind whether to buy a cat or a dog. So, I decide to decide when I go the pet store on Friday. My wife is a neuroscientist who hates cats almost as much as she hates healthy communication. While I’m sleeping, she inserts a device at a strategic point in my brain that prevents me from choosing a cat and nothing else. None the wiser, I go to the pet store on Friday and decide to get a dog, but entirely of my own accord.

Did I choose freely?

These examples evidence the mischief falling out of heuristic neglect in a stark way. My wife’s device only interferes with decision-making processes to prevent one undesirable output. If the output is desirable, it plays no role, suggesting that the hacked subject chose that output ‘freely,’ despite the inability to do otherwise. On the one hand, surgical intervention prevents the application of choice-talk to cat buying. Source-talk, after all, trumps choice-talk. But since surgical intervention only pertains to cat buying, dog buying seems, to some at least, to remain a valid subject of choice-talk. Source neglect remains unproblematic. The machinery of decision-making, in other words, can be ignored the way it’s always ignored in decision-making contexts. It remains irrelevant. Choice-talk machinery seems to remain applicable to this one fork, despite crashing when both forks are taken together.

For some philosophers, this suggests that choice isn’t a matter of being able to do otherwise, but of arising out of the proper process—a question of appropriate ‘sourcing.’ They presume that choice-talk and the corresponding intuitions still apply. If the capacity to do otherwise isn’t definitive of choice, then provenance must be: choice is entirely compatible with precursors, they argue, so long as those precursors are the proper ones. Crash. Down another interpretative rabbit-hole they go. Short any inkling of the limits imposed by the heuristic tools at their disposal—blind to their own cognitive capacities—all they can do is pursue the intuitions falling out of the misapplications of those tools. They remain trapped, in effect, downstream the heuristic confounds described above.

Here we can see the way philosophical parsing lets us map the boundaries of reliable choice-talk application. Frankfurt-style counterexamples, on this account, are best seen as cognitive versions of visual illusions, instances where we trip over the ecological limits of our cognitive capacities.

As with visual illusions, they reveal the fractionate, heuristic nature of the capacities employed. Unlike visual illusions, however, they are too low-dimensional to be readily identified as such. To make matters worse, the breakdown is socio-cognitive: perpetual disputation between individuals is the breakdown. This means that its status as a crash space is only visible by taking an ecological perspective. For interpretative partisans, however, the breakdown always belongs to the ‘other guy.’ Understanding the ecology of the breakdown becomes impossible.

The stark lesson here is that ‘free-will’ is a deliberative confound, what you get when you ponder the nature of choice-talk without accounting for heuristic neglect. Choice-talk itself is very real. With the interactional system it belongs to—intentional cognition more generally—it facilitates cooperative miracles on the communicative back of less than fifteen bit-per-second. Impressive. Gobsmacking, actually. We would be fools not to trust our socio-cognitive reflexes where they are applicable, which is to say, where neglecting sources solves more problems than it causes.

So, yah, sure, we make choices all the bloody time. At the same time, though, ‘What is the nature of choice?’ is a question that can only be answered ecologically, which is to say, via source-sensitive cognition. The nature of choice involves the systematic neglect of systems that must be manipulated nevertheless. Cues and correlations are compulsory. The nature of choice, in other words, obliterates our intellectual and phenomenological intuitions regarding choice. There’s just no such thing.

And this, I think it’s fair to say, is as disastrous as a natural fact can be. But should we be surprised? The thing to appreciate, I think, is the degree to which we should expect to find ourselves in precisely such a dilemma. The hard fact is that biocomplexity forced us to evolve source-insensitive ways to troubleshoot all organisms, ourselves included. The progressive nature of science, however, insures that biocomplexity will eventually succumb to source-sensitive cognition. So, what are the chances that two drastically different, evolutionarily segregated cognitive modes would be easily harmonized?

Perhaps this is a growing pain every intelligent, interstellar species suffers, the point where their ancestral socio-cognitive toolset begins to fail them. Maybe science strips exceptionalism from every advanced civilization in roughly the same way: first our exceptional position, then our exceptional origin, and lastly, our exceptional being.

Perhaps choice dies with the same inevitability as suns, choking on knowledge instead of iron.