The Cost of Consciousness

Tom Clark
Book Title: 
Book Author: 
Annaka Harris

A review of Annaka Harris's book Conscious.

The Cost of Consciousness


Does Phenomenology Have a Function?

If you’ve not encountered the so-called “hard problem” of consciousness, Annaka Harris’s Conscious – an accessible and engaging read – is a good place to start. Following philosopher Thomas Nagel, Harris pinpoints the tough question about conscious experience: why should it feel like something to instantiate certain brain processes? She then challenges common intuitions about consciousness, for instance that you can reliably tell when something or other is conscious, and that it serves a function in behavior control. The sensory feels of experience, e.g., being in pain or tasting a mango, are private episodes that need not have any overt behavior accompanying them, and it’s this privacy, the subjectivity of such states (as contrasted with their associated objectively accessible brain states), that makes finding reliable public criteria for consciousness problematic.[1] We know we are conscious, but how about those smart octopuses who bear only passing neural resemblance to us? Or bees? Or the flexibly intelligent robots of the future? This is one version of the problem of other minds.

Likewise, despite intuitions to the contrary, it isn’t immediately obvious what experience adds to behavior control that its associated neural states don’t already accomplish. In no scientific flow charts explaining pain-related behavior is there a pain box, only the various neurally-realized functions that when active result in the behavior and reports of pain.[2] What is pain itself doing that the neuro-muscular goings-on aren’t? You might respond: well, maybe pain just is those goings-on. But if so, then pain, being identical to them, can’t add to what they do, so we needn’t include it in the flow chart. This is a worry for those who think consciousness per se, as opposed to its neural correlates, obviously serves some naturally-selected function.

Harris does a nice job elucidating these concerns, and although she offers no definitive resolutions, tends toward epiphenomenalism about consciousness: it comes along for the ride, but serves little or no function.[3] This presents a puzzle since evolution doesn’t much traffic in useless add-ons for the fun of it, especially when they are as glucose-intensive as the brain processes associated with the kinds of  behavior that involve consciousness. As a leading philosopher of mind, Thomas Metzinger, puts it:

Every new virtual organ, every new sensory experience, every new conscious thought, had a metabolic price; it was costly to activate them, if only for a couple of seconds or minutes at a time. But since they paid for themselves in terms of additional glucose, and in terms of security, survival and procreation, they spread across populations and sustain themselves to this day.[4]

Metzinger supposes that experienced feels (conscious phenomenology) somehow convey a behavioral advantage (“they paid for themselves”) and thus are the point of having the metabolically expensive neural processes associated with them; if true, this would answer the question of why consciousness was naturally selected for. But Harris argues, correctly in my view, that it’s difficult to find a functional role for phenomenology per se, as in the pain example above (42, 65). Feels are real all right, but we needn’t appeal to them in explaining behavior, at least not from a scientific standpoint. (We appeal to them all the time in folk explanations, of course: I eat chocolate because it tastes good, right?). Evolution clearly selected for conscious processes – the neural processes associated with consciousness – since those processes support our capacities for learning, memory, anticipation, deliberation and complex and novel behavior, which obviously paid for themselves; but it didn’t obviously select for consciousness per se (79).

So, if not from evolution, from whence comes consciousness? Harris finds attractive the panpsychist possibility (so devotes many words to it) that consciousness isn’t essentially connected to naturally-selected conscious brain processes, but is instead a fundamental feature of reality at the micro-level, perhaps an intrinsic property of matter. This could explain why experience tags along (at no metabolic cost) with brain processes, which after all are constituted by micro-constituents such as molecules, atoms, quarks, etc.

The problem, however, is that as far as I know (and I’m happy to stand corrected), there is no empirical, observational evidence supporting the claim that experience, however simple and unstructured, attends the micro-constituents of brains as well as brains themselves. Were there such evidence, then it would be worth addressing, as does Harris at length, the so-called “combination problem” of how the (strictly hypothetical) informationally non-specific and abstract proto-consciousness of, say, quarks gets amalgamated into the specific, highly informational, concrete, multi-modal phenomenal gestalt of human brain-based experience. But until there’s at least a smidgen of empirical support for the existence of micro-level experience, we can safely leave panpsychism aside as a speculative pre-theoretical hunch.

Why this dalliance with panpsychism? Harris mainly touts its simplicity as scientifically virtuous (67-9), but simplicity on its own doesn’t get us very far. She may not have sufficiently considered the explanatory potential of neuroscientific hypotheses concerning consciousness, to which she gives short shrift in this volume. Commenting on research on the neural correlates of consciousness conducted by Christof Koch and the late Francis Crick, she says:

While useful and interesting, this type of research is, once again, limited. It increases our knowledge of the brain and our human experience, but it can’t tell us anything about what consciousness is in the first place, nor does it help us understand whether or not other types of systems, animate or inanimate, could be experiencing it.  (74, original emphasis)

This seems too pessimistic an assessment of how brain-based theories might shed light on the nature of consciousness and therefore what sorts of systems might harbor it.[5] Neuroscientist Anil Seth, who like Harris accepts that there is a real question about why sensory experience attends certain brain processes, counsels patience as the spade work gets done of connecting specific features of consciousness to specific neurally-realized functions.[6] And indeed, there’s considerable ongoing research on the sorts of behavior-control functions associated with consciousness, as contrasted with those that are not. Getting a fix on the specific nature of those functions, for instance their role in processing representations constrained by perceptual input (as in the now ascendant predictive coding paradigm), will likely winnow current hypotheses and suggest new ones. The most promising, as I see it, are those that take experience as carrying phenomenal (qualitative) representational content (information about the world, such as the apple being red) and then theorize why a system that represents reality (as we do in having brains) might necessitate the existence of such content for the system, and for it alone (experience being a private affair).[7]

Harris rightly rejects the currently fashionable illusionism about phenomenal content (71),[8] but leaves unmentioned or unexplored, perhaps to keep the book brief and accessible to the general reader, some of the major empirically-motivated explanatory contenders, for instance those of Thomas Metzinger (self-model theory of subjectivity), Jesse Prinz (attended intermediate representations), Giulio Tononi (integrated information theory) Stanislas Dehaene and Bernard Baars (global workspace), and Andy Clark (predictive coding). Panpsychist speculation about experience being the intrinsic character or nature of physical reality (if such a nature even exists) seems empirically bereft and premature compared to such work.

To be fair, Harris explicitly prioritizes the pursuit of information-processing, brain-based theories even as she defends a safe space for panpsychism (100). But to take panpsychism seriously seems to insufficiently respect the evidential requirement for a productive research hypothesis; the widespread reluctance among scientists to pursue it is, I think, well founded (what would they do?). That said, short of misallocated grant funding, there’s no material harm in speculating that in having experiences we are in touch with the root nature of reality. Would that things were that simple.

The metabolic cost of consciousness-as-experience (the target of the hard problem) is just that of the neurally-realized processes with which it is associated. If, as seems likely to me, conscious experience is not a causal product or further emergent property of those processes, but rather their representational content (which doesn’t require additional glucose to produce), then the solution to the hard problem lies in explaining why representational content sometimes, in certain specific cases, assumes a qualitative character and is available only to the system instantiating the relevant representations (see note 7). Getting clear about the informational goings-on of complex mind-systems like ourselves may well shed light on the entailment, should it exist, between representational functions and their associated phenomenology. This, it can be seen, is primarily a representationalist, not physicalist, research program.

Although she doesn’t consider this angle on consciousness, Harris nevertheless succeeds admirably in explaining and motivating the hard problem for those unfamiliar with it, and covers essential ground in consciousness studies. What is the status of the self? What determines our choices if not consciousness? What might be the minimal necessary content of a conscious state, for instance during meditation? What ethical issues are raised by the advent of possibly sentient machines? Readers, thus motivated by encountering a many-faceted but potentially solvable philo-scientific mystery, will want to explore further.[9] For my money there’s no more fascinating problem than that of why we’re conscious, and aside from my reservations about panpsychism, Conscious is a worthy introduction to it.

- Tom Clark, Naturalism.Org, June, 2019


[1] For a recent development on this, see the New York Times article on EEG tests for “covert consciousness.”

[2] For such flow charts, see Daniel Dennett’s Why you can't make a computer that feels pain.

[3] Even if consciousness doesn’t have a function, epiphenomenalism might mischaracterize its relation to behavior, see Respecting privacy: why consciousness isn’t even epiphenomenal.

[4] Thomas Metzinger, The Ego Tunnel, p. 58, reviewed here.

[5] For two recent books on evolution and the possible distribution of consciousness in the animal kingdom, see Ginsburg and Jablonka’s The Evolution of the Sensitive Soul: Learning and the Evolution of Consciousness and Feinberg and Mallat’s Consciousness Demystified.

[6] Anil Seth in conversation with Sam Harris on consciousness, here.

[7] For a representationalist approach to explaining consciousness, which draws extensively on the work of philosopher Thomas Metzinger, see T. Clark, The appearance of reality and T. Clark, “Locating consciousness: why experience can’t be objectified,” forthcoming in the Journal of Consciousness Studies.

[8] Illusionists argue there is really nothing it is like to feel pain, smell a rose, etc., etc. Rather we misrepresent some of our informational states as having phenomenal qualities. See Frankish, Dennett, et al., Illusionism as a Theory of Consciousness.

[9] My recommendations for further reading include The Ego Tunnel by Thomas Metzinger (reviewed here), Consciousness Demystified by Tod Feinberg and Jon Mallat, Susan Blackmore’s books on consciousness (reviewed here), and Philosophy in a Meaningless Life: A  System of Nihilism, Consciousness and Reality, by James Tartaglia. See also the consciousness page at Naturalism.Org for other papers and reviews.