The entry placing second (!!) in the 2016 Illusion of the Year competition, the Ambiguous Cylinder Illusion, blew up on Reddit for good reason. What you’re seeing below is an instance where visual guesswork arising from natural environmental frequencies have been cued ‘out of school.’ In this illusion, convex and concave curves trick the visual system into interpreting a ‘squircle’ as either a square or a circle—thus the dazzling images. Ambiguous cylinders provide dramatic illustration of a point Dennett makes many times in From Bacteria to Bach and Back: “One of the hallmarks of design by natural selection,” he writes, “is that it is full of bugs, in the computer programmer’s sense: design flaws that show up only under highly improbable conditions, conditions never encountered in the finite course of R&D that led to the design to date, and hence not yet patched or worked around by generations of tinkering” (83). The ‘bug’ exploited in this instance could be as much a matter of neural as natural selection, of course—perhaps, as with the Muller-Lyer illusion, individuals raised in certain environments are immune to this effect. But the upshot remains the same. By discovering ways to cue heuristic visual subsystems outside their adaptive problem ecologies, optical illusionists have developed a bona fide science bent on exploring what might be called ‘visual crash space.’
One of the ideas behind Three Pound Brain is to see traditional intentional philosophy as the unwitting exploration of metacognitive crash space. Philosophical reflection amounts to the application of metacognitive capacities adapted to trouble-shooting practical cognitive and communicative issues to theoretical problems. What Dennett calls ‘Cartesian gravity,’ in other words, has been my obsession for quite some time, and I think I have a fair amount of wisdom to share, especially when it comes to philosophical squircles, things that seem undeniable, yet nevertheless contradict our natural scientific understanding. Free will is perhaps the most famous of these squircles, but there’s really no end to them. The most pernicious squircle of all, I’m convinced, is the notion of intentionality, be it ‘derived’ or ‘original.’
On Heuristic Neglect Theory, Cartesian gravity boils down to metacognitive reflexes, the application of heuristic systems to questions they have no hope of answering absent any inkling of as much. The root of the difficulty lies in neglect, the way insensitivity to the limits of felicitous application results in various kinds of systematic errors (what might be seen as generalized versions of the WYSIATI effects discovered by Daniel Kahneman).
The centrality of neglect (understood as an insensitivity that escapes our sensitivity) underwrites my reference to the ‘Grand Inversion’ in the previous installment. As an ecological artifact, human cognition trivially possesses what might be called a neglect structure: we are blind to the vast bulk of the electromagnetic spectrum, for instance, because sensing things like gamma radiation, infrared, or radio waves paid no ancestral dividends. If fact, one can look at the sum of scientific instrumentation as mapping out human ‘insensitivity space,’ providing ingress into all those places our ancestral sensitivities simply could not take us. Neglect, in other words, allows us to quite literally invert our reflexive ways of comprehending comprehension, not only in a wholesale manner, but in a way entirely compatible with what Dennett calls, following Sellars, the scientific image.
Simply flipping our orientation in this way allows us to radically recharacterize Dennett’s project in From Bacteria to Bach and Back as a matter of implicitly mapping our human neglect structure by filling in all the naturalistic blanks. I say implicit because his approach remains primarily focused on what is neglected, rather than neglect considered in its own right. Despite this, Dennett is quite cognizant of the fact that he’s discussing a single phenomenon, albeit one he characterizes (thanks to Cartesian gravity!) in positive terms:
Darwin’s “strange inversion of reasoning” and Turing’s equally revolutionary inversion form aspects of a single discovery: competence without comprehension. Comprehension, far from being a god-like talent from which all design must flow, is an emergent effect of systems of uncomprehending competence… (75)
The problem with this approach is one that Dennett knows well: no matter how high you build your tower of natural processes, all you’ve managed to do, in an important sense, is recapitulate the mystery you’ve set out to solve. No matter how long you build your ramp, talk of indefinite thresholds and ‘emergent effects’ very quickly reveals you’re jumping the same old explanatory shark. In a sense, everyone in the know knows at least the moral of the story Dennett tells: competences stack into comprehension on any Darwinian account. The million-dollar question is how ‘all that’ manages to culminate in this…
Personally speaking, I’ve never had an experience quite like the one I had reading this book. Elation, realizing that one of the most celebrated minds in philosophy had (finally!) picked up on the same trail. Urgency, knowing I had to write a commentary, like, now. And then, at a certain point, wonder at the sense of knowing, quite precisely, what it was tantalizing his intuitions: the profound connection between his Darwinian commitments and his metaphilosophical hunches regarding Cartesian gravitation.
Heuristic Neglect Theory not only allows us to economize Dennett’s bottom-up saga of stacking competences, it also provides a way to theorize his top-down diagnosis of comprehension. It provides, in other words, the common explanatory framework required to understand this… in terms of ‘all that.’ No jumps. No sharks. Just one continuous natural story folding comprehension into competence (or better, behaviour).
What applies to human cognition applies to human metacognition—understood as the deliberative derivation of endogenous or exogenous behaviour via secondary (functionally distinct) access to one’s own endogenous or exogenous behaviour. As an ecological artifact, human metacognition is fractionate and heuristic, and radically so, given the complexity of the systems it solves. As such, it possesses its own neglect structure. Understanding this allows us to ‘reverse-engineer’ far more than Dennett suspects, insofar as it lets us hypothesize the kinds of blind spots we should expect to plague our attempts to theorize ourselves given the deliverances of philosophical reflection. It provides the theoretical basis, I think, for understanding philosophy as the cognitive psychological phenomenon that it is.
It’s a truism to say that the ability to cognize any system crucially depends on a cognitive system’s position relative to that system. But things get very interesting once we begin picking at the how and why. The rationality of geocentrism, for instance, is generally attributed to the fact that from our terrestrial perspective, the sky does all the moving. We remain, as far as we can tell, motionless. Why is motionlessness the default? Why not assume ignorance? Why not assume that the absence of information warranted ‘orbital agnosticism’? Basically, because we lacked the information to determine our lack of information.
Figure 1: It is a truism to state that where we find ourselves within a system determines our ability to cognize that system. ‘Frame neglect’ refers to our cognitive insensitivity, not only to our position within unknown systems, but to this insensitivity.
Figure 2: Thus, the problem posed by sufficiency, the automatic presumption that what we see is all there is. The ancients saw the stars comprising Orion as equidistant simply because they lacked the information and theory required to understand their actual position—because they had no way of knowing otherwise.
Figure 3: It is also a truism to state that the constitution of our cognitive capacities determines our ability to cognize systems. ‘Medial neglect’ refers to our cognitive insensitivity, not only to the constitution of our cognitive capacities, but to this insensitivity. We see, but absent any sensitivity to the machinery enabling sight.
Figure 4: Thus, once again, the problem posed by sufficiency. Our brain interprets ambiguous cylinders as magical squircles because it possesses no sensitivity to the kinds of heuristic mechanisms involved in processing visual information.
Generally speaking, we find these ‘no information otherwise’ justifications so intuitive that we just move on. We never ask how or why the absence of sensible movement cues reports of motionlessness. Plato need only tell us that his prisoners have been chained before shadows their whole lives and we get it, we understand that for them, shadows are everything. By merely conjuring an image, Plato secures our acknowledgment that we suffer a congenital form of frame neglect, a cognitive insensitivity to the limits of cognition that can strand us with fantastic (and so destructive) worldviews—and without our permission, no less. Despite the risk entailed, we neglect this form of neglect. Though industry and science are becoming ever more sensitive to the problems posed by the ‘unknown unknown,’ it remains the case that each of us at once understands the peril and presumes we’re the exception, the system apart from the systems about us. The motionless one.
Frame neglect, our insensitivity to the superordinate systems encompassing us, blinds us to our position within those systems. As a result, we have no choice but to take those positions for granted. This renders our cognitive orientations implicit, immune to deliberative revision and so persistent (as well as vulnerable to manipulation). Frame neglect, in other words, explains why bent orientations stay bent, why we suffer the cognitive inertia we do. More importantly, it highlights what might be called default sufficiency, the congenital presumption of implicit cognitive adequacy. We were in no position to cognize our position relative the heavens, and yet we nevertheless assumed that we were simply because we were in no position to cognize the inadequacy of our position.
Why is sufficiency the presumptive default? The stacking of ‘competences’ so brilliantly described by Dennett requires that every process ‘do its part’: sufficiency, you could say, is the default presumption of any biological system, so far as its component systems turn upon the iterative behaviour of other component systems. Dennett broaches the notion, albeit implicitly, via the example of asking someone to report on a nearby house via cell phone:
Seeing is believing, or something like that. We tacitly take the unknown pathways between his open eyes and speaking lips to be secure, just like the requisite activity in the pathways in the cell towers between his phone and ours. We’re not curious on the occasion about how telephones work; we take them for granted. We also don’t scratch our heads in bafflement over how he can just open his eyes and then answer questions with high reliability about what is positioned in front of him in the light, because we can all do it (those of us who are not blind). 348-349
Sufficiency is the default. We inherit our position, our basic cognitive orientation, because it sufficed to solve the kinds of high-frequency and/or high impact problems faced by our ancestors. This explains why unprecedented circumstances generate the kinds of problems they do: it’s always an open question whether our basic cognitive orientation will suffice when confronted with a novel problem.
When it comes to vision, for instance, we possess a wide range of ways to estimate sufficiency and so can adapt our behaviour to a variety of lighting conditions, waving our hand in fog, peering against glares, and so on. Darkness in particular demonstrates how the lack of information requires information, lest it ‘fall off the radar’ in the profound sense entailed by neglect. So even though we possess myriad ways to vet visual information, squircles possess no precedent and so no warning, the sufficiency of the information available is taken for granted, and we suffer the ambiguous cylinder illusion. Our cognitive ecology plays a functional role in the efficacy of our heuristic applications—all of them.
From this a great deal follows. Retasking some system of competences always runs the risk of systematic deception on the one hand, where unprecedented circumstances strand us with false solutions (as with the millennia-long ontological dualism of the terrestrial and the celestial), and dumbfounding on the other, where unprecedented circumstances crash some apparently sufficient application in subsequently detectable ways, such as ambiguous for human visual systems, or the problem of determinism for undergraduate students.
To the extent that ‘philosophical reflection’ turns on the novel application of preexisting metacognitive resources, it almost certainly runs afoul instances of systematic deception and dumbfounding. Retasked metacognitive channels and resources, we can be assured, would report as sufficient, simply because our capacity to intuit insufficiency would be the product of ancestral, which is to say, practical, applications. How could information and capacity geared to catching our tongue in social situations, assessing what we think we saw, rehearsing how to explain some disaster, and so on hope to leverage theoretical insights into the fundamental nature of cognition and experience? It can’t, no more than auditory cognition, say, could hope to solve the origin of the universe. But even more problematically, it has no hope of intuiting this fundamental inability. Once removed from the vacuum of ecological ignorance, the unreliability of ‘philosophical reflection,’ its capacity to both dumbfound and to systematically deceive, becomes exactly what we should expect.
This follows, I think, on any plausible empirical account of human metacognition. I’ve been asking interlocutors to provide me a more plausible account for years now, but they always manage to lose sight of the question somehow.
On the availability side, we should expect the confusion of task-insufficient information with task-sufficient information. On the capacity side, we should expect the confusion of task-insufficient applications with task-sufficient applications. And this is basically what Dennett’s ‘Cartesian gravity’ amounts to, the reflexive deliberative metacognitive tendency to confuse scraps with banquets and hammers with swiss-army knives.
But the subtleties secondary to these reflexes can be difficult to grasp, at least at first. Sufficiency means that decreases in dimensionality, the absence of kinds and quantities of information, simply cannot be cognized as such. Just over two years ago I suffered a retinal tear, which although successfully repaired, left me with a fair amount of debris in my right eye (‘floaters,’ as they call them, which can be quite distracting if you spend as much time staring at white screens as I do). Last autumn I noticed I had developed a ‘crimp’ in my right eye’s field of vision: apparently some debris had become attached to my fovea, a mass that accumulated as I was passed from doctor to doctor and thence to the surgeon. I found myself with my own, entirely private visual illusion: the occluded retinal cells were snipped out of my visual field altogether, mangling everything I tried to focus on with my right eye. The centre of every word I looked at would be pinched into oblivion, leaving only the beginning and ending characters mashed together. Faces became positively demonic—to the point where I began developing a Popeye squint for equanimity’s sake. The world had become a grand bi-stable image: things were fine when my left eye predominated, but then for whatever reason, click, my friends and family would be eyeless heads of hair. Human squircles.
My visual centres simply neglected the missing information, and muddled along assuming the sufficiency of the information that was available. I understood the insufficiency of what I was seeing. I knew the prisoners were there, chained in their particular neural cave with their own particular shadows, but I had no way of passing that information upstream—the best I could do was manage the downstream consequences.
But what happens when we have no way of intuiting information loss? What happens when our capacity to deliberate and report finds itself chained ‘with no information otherwise’? Well, given sufficiency, it stands to reason that what metacognition cannot distinguish we will report as same, that what it cannot vet we will report as accurate, that what it cannot swap we will report inescapable, and that what it cannot source we will report as sourceless, and so on. The dimensions of information occluded, in other words, depend entirely on what we happen to be reporting. If we ponder the proximate sources of our experiences, they will strike us as sourceless. If we ponder the composition of our experiences, they will strike us simple. Why? Because human metacognition not only failed to evolve the extraordinary ability to theoretically source or analyze human experience, it failed to evolve the ability to intuit this deficit. And so, we find ourselves stranded with squircles, our own personal paradox (illusion) of ourselves, of what it is fundamentally like to be ‘me.’
Dialectically, it’s important to note how this consequence of the Grand Inversion overturns the traditional explanatory burden when it comes to conscious experience. Since it takes more metacognitive access and capacity, not less, to discern things like disunity and provenance, the question Heuristic Neglect Theory asks of the phenomenologist is, “Yes, but how could you report otherwise?” Why think the intuition of apperceptive unity (just for instance) is anything more than a metacognitive cousin of the flicker-fusion you’re experiencing staring at the screen this very instant?
Given the wildly heuristic nature of our metacognitive capacities, we should expect to possess the capacity to discriminate only what our ancestors needed to discriminate, and precious little else. So, then, how could we intuit anything but apperceptive unity? Left with a choice between affirming a low-dimensional exception to nature on the basis of an empirically implausible metacognitive capacity, and a low-dimensional artifact of the very kind we might expect given an empirically plausible metacognitive account, there really is no contest.
And the list goes on and on. Why think intuitions of ‘self-identity’ possess anything more than the information required to resolve practical, ancestral issues involving identification?
One can think of countless philosophical accounts of the ‘first-person’ as the product of metacognitive ‘neglect origami,’ the way sufficiency precludes intuiting the radical insufficiency of the typically scant dimensions of information available. If geocentrism is the default simply for the way our peripheral position in the solar system precludes intuiting our position as peripheral, then ‘noocentrism’ is the default for the way our peripheral position vis a vis ourselves precludes intuiting our position as peripheral. The same way astrophysical ignorance renders the terrestrial the apparently immovable anchor of celestial motion, metacognitive neglect renders the first-person the apparently transcendent anchor of third-person nature. In this sense, I think, ‘gravity’ is a well-chosen metaphor to express the impact of metacognitive neglect upon the philosophical imagination: metacognitive neglect, like gravity, isn’t so much a discrete force as a structural feature, something internal to the architecture of philosophical reflection. Given it, humanity was all but doomed to wallow in self-congratulatory cartoons once literacy enabled regimented inquiry into its own nature. If we’re not the centres of the universe, then surely we’re the centre of our knowledge, our projects, our communities—ourselves.
Figure 5: The retasking of deliberative metacognition is not unlike discovering something practical—such as ‘self’ (or in this case, Brian’s sandal)—in apparently exceptional, because informationally impoverished, circumstances.
Figure 6: We attempt to interpret this practical deliverance in light of these exceptional circumstances.
Figure 7: Given neglect, we presume the practical deliverance theoretically sufficient, and so ascribe it singular significance.
Figure 8: We transform ‘self’ into a fetish, something both self-sustaining and exceptional. A squircle.
Of all the metacognitive misapplications confounding traditional interpretations of cognition and experience, Dennett homes in on the one responsible for perhaps the most theoretical mischief in the form of Hume’s ‘strange inversion of reasoning’ (354-358), where the problem, as we saw in the previous post, lies in mistaking the ‘intentional object’ of the red stripe illusion for the cause of the illusion. Hume, recall, notes our curious propensity to confuse mental determinations for environmental determinations, to impute something belonging to this… to ‘all that.’ Dennett notes that the problem lies in the application: normally, this ‘confusion’ works remarkably well; it’s only in abnormal circumstances, like those belonging to the red stripe illusion, where this otherwise efficacious cognitive reflex leads us astray.
The first thing to note about this cognitive reflex is the obvious way it allows us to neglect the actual machinery of our environmental relations. Hume’s inversion, in other words, calls attention to the radically heuristic nature of so-called intentional thinking. Given the general sufficiency of all the processes mediating our environmental relationships, we need not cognize them to cognize those relationships, we can take them for granted, which is a good thing, because their complexity (the complexity cognitive science is just now surmounting) necessitates they remain opaque. ‘Opaque,’ in this instance, means heuristically neglected, the fact that all the mad dimensionalities belonging to our actual cognitive relationships appear nowhere in cognition, not even as something missing. What does appear? Well, as Dennett himself would say, only what’s needed to resolve practical ancestral problems.
Reporting environments economically entails taking as much for granted as possible. So long as the machinery you and I use to supervise and revise our environmental orientations is similar enough, we can ignore each other’s actual relationships in communication, focusing instead on discrepancies and how to optimize them. This is why we narrate only those things most prone to vary—environmentally and neurally sourced information prone to facilitate reproduction—and remain utterly oblivious to the all the things that go without saying, the deep information environment plumbed by cognitive science. The commonality of our communicative and cognitive apparatuses, not to mention their astronomical complexity, assures that we will suffer what might be called, medial neglect, congenital blindness to the high-dimensional systems enabling communication and cognition. “All the subpersonal, neural-level activity is where the actual causal interactions happen that provide your cognitive powers, but all “you” have access to is the results” (348).
From Bacteria to Bach and Back is filled with implicit references to medial neglect. “Our access to our own thinking, and especially to the causation and dynamics of its subpersonal parts, is really no better than our access to our digestive processes,” Dennett writes; “we have to rely on the rather narrow and heavily edited channel that responds to our incessant curiosity with user-friendly deliverances, only one step closer to the real me than the access to the real me that is enjoyed by my family and friends” (346).
Given sufficiency, “[t]he relative accessibility and familiarity of the outer part of the process of telling people what we can see—we know our eyes have to be open, and focused, and we have to attend, and there has to be light—conceals from us the other blank from the perspective of introspection or simple self-examination of the rest of the process” (349). The ‘outer part of the process,’ in other words, is all that we need.
Medial neglect may be both necessary and economical, but it remains an incredibly risky bet to make given the perversity of circumstance and the radical interdependency characterizing human communities. The most frequent and important discrepancies will be environmental discrepancies, those which, given otherwise convergent orientations (the same physiology, location, and training), can be communicated absent medial information, difference making differences geared to the enabling axis of communication and cognition. Such discrepancies can be resolved while remaining almost entirely ‘performance blind.’ All I need do is ‘trust’ your communication and cognition, build upon it the same blind way I build upon my own. You cry, ‘Wolf!’ and I run for the shotgun: our orientations converge.
But as my example implies, things are not always so simple. Say you and I report seeing two different birds, a vulture versus an albatross, in circumstances where such a determination potentially matters—looking for a lost hunting party, say. An endless number of medial confounds could possibly explain our sudden disagreement. Perhaps I have bad eyesight, or I think albatrosses are black, or I’m blinded by the glare of the sun, or I’m suffering schizophrenia, or I’m drunk, or I’m just sick and tired of you being right all the time, or I’m teasing you out of boredom, or more insidiously, I’m responsible for the loss of the hunting party, and want to prevent you from finding the scene of my crime.
There’s no question that, despite medial neglect, certain forms of access and capacity regarding the enabling dimension of cognition and communication could provide much in the way of problem resolution. Given the stupendous complexity of the systems involved, however, it follows that any capacity to accommodate medial factors will be heuristic in the extreme. This means that our cognitive capacity to flag/troubleshoot issues of sufficiency will be retail, fractionate, geared to different kinds of high-impact, high-frequency problems. And the simplest solution, the highest priority reflex, will be to ignore the medial altogether. If our search party includes a third soul who also reports seeing a vulture, for instance, I’ll just be ‘wrong’ for ‘reasons’ that may or not be determined afterward.
The fact of medial neglect, in other words, underwrites what might be called an environmentalization heuristic, the reflexive tendency to ‘blame’ the environment first.
When you attempt to tell us about what is happening in your experience, you ineluctably slide into a metaphorical idiom simply because you have no deeper, truer, more accurate knowledge of what was going on inside you. You cushion your ignorance with a false—but deeply tempting—model: you simply reproduce, with some hand waving and apologies, your everyday model of how you know about what is going on outside you. 348
Because that’s typically all that you need. Dennett’s hierarchical mountain of competences is welded together by default sufficiency, the blind mechanical reliance of one system upon other systems. Communicative competences not only exploit this mechanical reliance, they extend it, opening entirely novel ecosystems leveraging convergent orientation, brute environmental parallels and physiological isomorphisms, to resolve discrepancies. So long as those discrepancies are resolved, medial factors potentially impinging on sufficiency can be entirely ignored, and so will be ignored. Communications will be ‘right’ or ‘wrong,’ ‘true’ or ‘false.’ We remain as blind to the sources of our cognitive capacities as circumstances allow us to be. And we remain blind to this blindness as well.
When I say from the peak of my particular competence mountain, “Albatross…” and you turn to me in perplexity, and say from the peak of your competence mountain, “What the hell are you talking about?” your instance of ‘about-talk’ is geared to the resolution of a discrepancy between our otherwise implicitly convergent systems. This is what it’s doing. The idea that it reveals an exceptional kind of relationship, ‘aboutness,’ spanning the void between ‘albatross’ here and albatrosses out there is a metacognitive artifact, a kind of squircle. For one, the apparent void is jam-packed with enabling competences—vast networks of welded together by sufficiency. Medial neglect merely dupes metacognition into presuming otherwise, into thinking the apparently miraculous covariance (the product of vast histories of natural and neural selection) between ‘sign’ (here) and ‘signified’ (out there) is indeed some kind of miracle.
Philosophers dwell among general descriptions and explanations: this is why they have difficulty appreciating that naïveté generally consists in having no ‘image,’ no ‘view,’ regarding this or that domain. They habitually overlook the oxymoronic implication of attaching any ‘ism’ to the term ‘naïve.’ Instances of ‘about-talk’ do not implicitly presume ‘intentionality’ even in some naïve, mistaken sense. We are not born ‘naïve intentionalists’ (any more than we’re ‘naïve realists’). We just use meaning talk to solve what kinds of problems we can where we can. Granted, our shared metacognitive shortcomings lead us, given different canons of interrogation, into asserting this or that interpretation of ‘intentionality,’ popular or scholastic. We’re all prone to see squircles when prompted to peer into our souls.
So, when someone asks, “Where does causality lie?” we just point to where we can see it, out there on the billiard table. After all, where the hell else would it be (given medial neglect)? This is why dogmatism comes first in the order of philosophical complication, why Kant comes after Descartes. It takes time and no little ingenuity to frame plausible alternatives of this ‘elsewhere.’ And this is the significance of Hume’s inversion to Cartesian gravity: the reflexive sufficiency of whatever happens to be available, a sufficiency that may or may not obtain given the kinds of problem posed. The issue has nothing to do with confusing normal versus abnormal attributions of causal efficacy to intentional objects, because, for one, there’s just no such thing as ‘intentional objects,’ and for another, ‘intentional object-talk’ generates far more problems than it solves.
Of course, it doesn’t seem that way to Dennett whilst attempting to solve for Cartesian gravity, but only because, short theoretical thematizations of neglect and sufficiency, he lacks any real purchase on the problem of explaining the tendency to insist (as Tom Clark does) on the reality of the illusion. As a result, he finds himself in the strange position of embracing the sufficiency of intentionality in certain circumstances to counter the reflexive tendency to assume the sufficiency of phenomenality in other circumstances—of using one squircle, in effect, to overcome another. And this is what renders him eminently vulnerable to readings like Clark’s, which turns on Dennett’s avowal of intentional squircles to leverage, on pain of inconsistency, his commitment to phenomenal squircles. This problem vanishes once we recognize ourselves for the ambiguous cylinders we have always been. Showing as much, however, will require one final installment.