I’ve recently been re-reading A. David Milner and Melvyn Goodale’s The Visual Brain in Action, as part of a project on the conscious control of behavior (concerning which I’ll hopefully have something to post rather soon). For those who aren’t familiar with the book, it presents a detailed empirical case for Milner and Goodale’s influential interpretation of the “two visual systems hypothesis”, according to which the human brain contains two functionally distinct visual streams, one responsible for providing the contents of visual consciousness and the other responsible for visuomotor control, which is supposed to be an entirely unconscious affair. In the first chapter, Milner and Goodale start to make their case by appeal to evolutionary considerations: they argue that vision must have evolved to control behavior, and that the existence of an unconscious primitive visuomotor module makes sense in this light. In this context, we get remarks like the following (p. 19):
“… it is a truism that all visual processing systems ultimately serve to guide behavior, otherwise they would not have evolved at all.”
And then later on (now in the epilogue to the second edition of the book, at p. 221), they write:
“… unless percepts can somehow be translated into action, they will have no consequences for the individual possessing them. Nor indeed would the brain systems that generate them ever have evolved. In other words, perception per se is not what the ventral stream is there for: perception is not an end in itself, in biological terms, but rather a means to an end.”
Claims like these seem to me to rest on several misunderstandings of the nature of evolutionary explanations, which are worth dwelling on only because they are exceedingly common. In the first place, it simply is not true that everything that “evolved” must “ultimately serve” to do anything at all: this is the whole point of Gould and Lewontin’s idea of a “spandrel”, or a non-adaptive trait that is the byproduct of some other trait that was selected for. Secondly, even if a trait once increased the fitness of an organism or population by guiding behavior in one way, it doesn’t follow that that trait still has this as its end: to give a toy example, suppose that we have taste buds because they enabled our ancestors to distinguish food that was rotten or unripe from food that was good to eat. (I’ve got no idea whether this is true, but that doesn’t matter.) Clearly it doesn’t follow that taste buds still help us do this: these days we’ve got better ways to tell whether food is good for eating, and our sense of taste is mostly good for helping us enjoy our meals, and tell the difference between Shiraz and Cabernet. More generally, even if you believe (as I do) that there’s explanatory value in teleological notions like “means” and “end in itself”, it needn’t be that the “ultimate purpose” of every important human characteristic is one that evolution would have cared about: for example, you might think that the most significant thing that conscious perception does for us is allow us to contemplate the mysteries of the universe, experience beauty, or become morally virtuous, each of which are “consequences” that organisms like us tend to care about even though they’re not especially adaptive.
Or is there something I’m missing?
Of course none of this shows that visual consciousness doesn’t influence human behavior (it does; and in fact I believe it has much more of a role in this than Milner and Goodale think), or that didn’t improve human fitness because of this. And there’s very good reason to believe that the conscious ventral stream was a later adaptation than the unconscious dorsal one, though the real evidence for this is the fact that the latter lies in a phylogenetically older part of the brain than the former. But there’s more lost than gained when we try to argue for these claims with teleologically-laden “Just-So” stories of the sort that I’ve quoted above.
P.S. Hi, this is my first post here. I’m a recent-ish Berkeley graduate who now teaches at a liberal arts college in Maryland. Here is my home page, and thanks to Gualtiero for the invitation to post here, which I’ve taken far too long to get around to doing.
Hi John,
Thanks for the post, and welcome to Brains!
I would say that the addition of “in biological terms” mostly saves them here. It’s clear when they’re talking about purposes, they’re talking about biological functions rather than “what we care about” as you are in your examples.
Of course, there still is the spandrel problem, but that seems vastly unlikely in the case of a whole visual module. (I say this also bearing in mind that what matters to function is not just the original reason why the mechanism proliferated in the population, but also why it continued and continues to be maintained in the population.)
So I guess I’m mostly OK with their “truisms” actually being truistic, biologically construed. Does that sound OK to you?
I’m not sure that that really helps them, Dan. First, I agree with you that it’s massively unlikely that conscious vision is a spandrel, but that shouldn’t be confused with the “truism” (which is not a truism, because it’s false) that the only things that evolved are things that serve to guide behavior. Second, it doesn’t seem to me that even in the “biological sense” the “purpose” of something has to be the same as what it was selected for: something might once have been adaptive because it did something that it no longer does (or does much) at all; and as a species changes it’s possible that certain of its characteristics might come to serve new purposes that are different from the ones they were selected for (if indeed they were “selected for” at all). By my lights, we can tell that conscious vision has the (biological) function of guiding behavior simply by noting how clumsy we’d be without it.
Hi John,
I appreciate your concern for precision but I think Dan is about right. When you say that we can tell that conscious vision has the function of guiding behavior by noting how clumsy we’d be without it, I think you getting quite close to making Milner and Goodale’s point in a different way. All you need is the plausible assumption that evolution plays a role in developing, shaping, and preserving important traits that make such an important difference.
M&G may be somewhat loose in their appeal to evolution, but in their context, it seems to me that they mean something fairly innocuous, along the lines of: almost certainly evolutionary processes played an important role in developing, shaping, and preserving the visual system because the visual system contributes to guiding behavior.
(And they surely don’t mean that the only things that evolve are things that guide behavior; they just mean that there is no other plausible biological function of the visual system that explains why evolution would develop, shape and preserve such a system.)
If that’s the claim they’re making, then I’ve got no objection. But I think they mean to be arguing the other way: it’s because we know that conscious vision evolved that we can tell that it “ultimately serve[s] to guide behavior”.
Hi John, welcome to the party.
I am not sure how much it adds, but here is my interpretation of the above M and G quote:
They are assuming that the visual system is an adaptation. The question then becomes, what it adapted for. One might think that they only reason we have ANY of our external sense modalities is to provide the monitoring side of the (broadly construed) monitoring and control we do (i.e. representing and intervening in the external world.) So, if one had this general picture of why we have sense modalities at all, then it is a “truism”.
This does help. But it still leaves unaddressed the second of the worries I raised, which is that even if if a given characteristic was adaptive because it did do x, it doesn’t follow that it still does that same thing for us (i.e., has that as its primary “end” or “purpose”).
Ultimately my complaint here is with presenting what’s really a very commonsense and empirically well-supported position (as each of these comments has observed) all dressed-up in an evolutionary story that (1) doesn’t provide much further support for it and (2) seems to me to run close to misconstruing the way that evolutionary explanation works (in particular, regarding natural selection as more purpose-driven and -driving than it actually is).
Hi John,
OK, here’s another attempted gloss. Let me know if/where you get off the boat.
When we’re aiming to understand an adapted biological device, it’s crucial to know what it was selected for. For example, Mark Changizi has convincingly argued that the reason that our three cones have the sensitivities they do is to detect changes in blood flow & oxygenation in exposed skin, in order to detect emotions. This adds a depth of understanding that is impossible to achieve without a selectional orientation – e.g. to know only *what* the cone sensitivities are, rather than why; or to know that they’re useful for discerning cloud shapes and wine vintages as well as blushes – but to be ignorant of the differing explanatory depth of these capacities. This point seems to apply whatever the facts are about what we *now* use the strange asymmetry in our cone sensitivities for (= effects that we now conceptually classify as useful or important? Not sure what else you could have in mind…)
Similarly, to understand the ventral stream, it’s crucial to know why it is the way it is, why it was selected to be that way. It’s a psychological system, so ultimately it must exert its selectionally relevant effects via behaviour. Some of those uses will be obvious and unsurprising, while some may be surprising (like the cone sensitivities are surprising). But what’s clear is that it’s crucial to understand the evolutionary *why*. (This even goes for spandrels: it’s crucial to know *whether* some effect is a spandrel.)
So while it may be useful & interesting to understand how a device achieves effects it wasn’t selected to achieve, there’s a big explanatory asymmetry between the selected effects and the others. And insofar as M and G can show that the effects they talk about were selected for, that does indeed lend support to their theory as having identified the fundamental effects – the primary functions.
Dan,
I absolutely see the value in the kind of evolutionary explanation you cite here. But here are two ways in which what M&G are saying strikes me as importantly different:
1) Changizi’s claim is much more detailed (it doesn’t just say that our cones are the way they are in order to guide behavior), and as such it rests on a different sort of evidence than the “truism” that whatever evolved, evolved because it somehow affects behavior. I would be very interested in a correspondingly detailed look into why the ventral system is the way it is, but the kind of throwaway remarks I cite (which are all that M&G really have to say about evolution and the ventral stream; their stuff on the dorsal stream is much better) don’t amount to that.
2) It seems to me that there’s an important distinction to be drawn between understanding a biological device and identifying its “ultimate purpose” or “end”. Surely these tasks aren’t unrelated, but e.g. I imagine you’d allow that it doesn’t follow from Changizi’s position that the detection of emotions is what our cones are “there for”, i.e. that that’s the most important thing they do for us now. Rather, if he is right then it is indeed true that we have the cones we do (i.e., that our cones are the way they are) because they did help us to detect emotions, but it could still be the case that now they do many things other than that – and perhaps these roles are much more important (i.e., contribute more to our survival, do more to enhance our lives, etc.) than the ones they were originally selected for. Moreover, as I said in my post it’s quite possible that a characteristic could have been adaptive because it did something for us that it no longer does at all, and was simply taken over by some new function. It’s this kind of idea that I was driving at with my toy example of using smell to detect ripe from rotten fruit, though the example you give here seems to me to work much better.
John-
Fine first post. My sense is that natural science for the last several centuries and evolution in particular have been about eliminating teleology from discussions about the world. Planets revolve around stars because they do, with their motion described by a few relatively simple laws. Survivors survive because they survive. I think that things get confusing because it is almost impossible to speak in non-teleological terms, or non-anthropomorphic terms, even when you are really speaking loosely, metaphorically. So we sort of assume that genes want to survive and pass themselves on, and that we evolve traits in order to guide behavior. Sometimes the fact that this is loose and metaphorical gets lost along the way somewhere. So yes, you are right, if we are going to be evolutionary puritans, we should eliminate teleological terms from our explanations. But I would imagine that M&G would say that of course nothing really evolves “in order to” do anything, but over eons, nature is so good at structures and mechanisms that are perfectly suited to survival, it doesn’t hurt to speak of nature “designing” these things, and doing so “in order to” survive. Am I being too charitable?
As to spandrels, I’ve often thought that spandrel = “data that doesn’t fit neatly into my theory”. A little less snidely, nature does seem to have an amazing ability to repurpose things (bones for structure becoming conduits for sound, scales becoming feathers that enable flight, etc.).
And this whole discussion gets into some big arguments about consciousness as epiphenomenon, since zombies would be just as evolutionarily fit as us, but I won’t go there right now . . .
-John Gregg
https://home.comcast.net/~johnrgregg
Thanks, Gregg. I actually don’t share this negative view of teleological explanations, though. I don’t think there’s anything problematically spooky about saying that the (“biological”) purpose of A in species B is to C: all that this means is that by C-ing, A enables members of B to survive, and to do B-ish sorts of things. This kind of talk is all over the place in biology, and I’m not sure that biology could even *be* biology if it simply got rid of it. (Whether you think this shows that biology is somehow second-rate will depend on your view of reductionism, of causal explanation, etc.) But it has nothing to do with what things “want” to do; an organism could very well not want to survive or propagate its species, but still its heart would be for pumping blood, and we’d be able to see that something had gone wrong (not in the moral sense, but just in the sense of its being an unhealthy member of its species) if its heart stopped doing this.
Your examples of “repurposing” are very helpful, though – a much better illustration than my toy ones of why “A evolved (/was ‘selected for’) to do B” doesn’t analyze “The purpose of A is to B”, which is what it seems to me that M&G were assuming.
P.S. I meant ‘John’, of course, not ‘Gregg’.
Re: 1 – OK!
Re: 2 – So there are two ways of thinking about the purpose of a biological device: 1) its proper function (selectional), and 2) the most important thing(s) it does for us now. I’m saying that understanding a biological device depends almost exclusively on (1), whereas you want to say it can also depend a lot on (2).
With respect to (2), I take it that you want to say there can be “important roles” that are not selected for; and since they’re important, they’re essential to understanding the device. But one way you have of characterizing the importance of these “non-selected” roles – namely as *now* contributing to our survival – should really be included in (1). If it *now* contributes to our survival, it probably did so in the recent past as well, thus (in part) explaining the prevalence of the trait in the population. In other words, the relevant aspect of the device was selected for performing that role. It’s a mistake to confuse proper function with *originally* selected-for role.
This “relevant aspect” bit is important too – no, I wouldn’t say that our cones are there in order to allow us to detect emotions; but I *would* say that they have the particular distribution of sensitivities they do so that we can detect emotions. (And they still perform that role for us, in spades.) The cones also have many other overlapping functions in virtue of the fact that they’re used – where that use has been selectionally orchestrated – by many other systems downstream, e.g. cortical systems.
So I think it will turn out that almost all of the roles you’d classify as important will turn out to have been selected for. Coupled with the looseness of the term “understanding,” I suspect we don’t disagree much, in the end. (As I said, I’m happy to allow that “understanding” can be enhanced by learning how an unselected role is performed, if that role enhances our lives. It’s interesting to learn how it is that aspartame tastes sweet. As this example illustrates, though, these unselected roles will often poach on the selected ones – indeed, we find them important only *because* they’re poachers. So maybe the selected role ends up being primary anyway.)
I’m glad to see this convergence, Dan, and I absolutely agree about the importance (indeed, perhaps the primary importance) of evolutionary explanations for our understanding of why things are the way they are, even if accounts of purpose and proper function aren’t reducible to them.
Great Info! Every once in a while I find something interesting.