The linguist James R Hurford recently wrote a popular introduction to the evolution of language ‘The Origins of Language: A slim guide’. As popular introductions go the book is one of the best out there. Hurford manages to deal things like the FOXP2 gene in a nuanced manner, likewise when discussing the KE family who supposedly have language specific impairments but spared intelligence he deals with the evidence in a balanced manner. He also discusses in detail things like the importance of imitation and social learning through triangulation (topics sometimes ignored by linguists) this approach is at odds with the highly polemical approach of people like Pinker and Chomsky who go out of their way to pretend that all the evidence supports them and none supports theorists of a different persuasion. Overall, I think that Hurford’s book is one of the best popular level introductions to the topic out there. However I did have some difficulties chapter 4 of the book where Hurford discusses animal concepts as precursors to human language development.

Hurford thinks that the correct way to start with discussing simple concepts is to discuss simple examples of animals classifying things in their environment. A necessary (but not sufficient) condition of a creature having a concept is their being able to classify things in their environment. A lot of reflex behaviours of creatures satisfy this criterion. He notes that Frogs have a reflex that ensures they only swat at certain objects moving in certain ways. So the frog’s brain is directly connected to the tongue and it swats when objects within a certain proximity move a certain way. All of this is common place and has been widely discussed in the literature since Lettvin et al. ‘What the Frog’s Eye tells the Frog’s Brain’ (1959).

Hurtford then goes on to make the following claims about the frog swatting the fly. For the Frog the class of events (moving black objects) has meaning or significance, but it would be a mistake to ask what the moving insect was significant of to the frog. He asserts that the moving insect is not a symbol of anything to the frog. But he argues the frog’s brain is partitioned to organise things into flies and not flies. For Hurford these frogs have internal mental representations, of classes of objects, events, and states of affairs. He is careful to note that the frog’s perception of a fly goes straight to his motor mechanisms and the response is not stored in memory. So in this sense he argues the frog does not have a full concept of a fly. The frog can though form a percept of the fly and this percept has significance for the frog. And the ability of the frog to form significant percepts is for Hurtford the start on the road to developing fully fledged concepts (‘The Origins of Language’ p. 62).

Now when one hears that a frog has a percept of a fly, one thinks of the frog perceiving a meaningful image of a fly. This image is meaningful because it moves a certain way, has a certain shape and size etc. However Hurford warns that this is not how he intends us to interpret his claims. He assures us that he does not believe that there are pictures in the head which the frog inspects because pictures are two static and are made of the wrong stuff. However his explanation from this point in interms of networks of neurons connected to each other while mainly accurate (though simplified), seems to move from the personal level of explanation to the sub-personal level without offering even a cursory defence of doing so.

At the sub-personal level we can talk of the brain interpreting particular patterns in its environment triggering certain reflexes like fly swatting. This is similar to the way we can talk about a thermostat interpreting particular patterns in its environment which cause the thermostat to behave in certain ways. But talk of meaningful percepts is out of place in both cases. With this in mind one wonders why Hurford speaks of the frog having meaningful percepts, this implies conscious awareness of the black dots, however he has not provided evidence to support this picture. Likewise when he speaks of frogs distinguishing between flies and non-flies this gives a confused picture. The frog is distinguishing between dots of a certain size and motion but as it has no concept of a fly, talk of the frog distinguishing between flies and non-flies is way off base. He also argues that the frog is not inspecting pictures in his mind (because pictures are made of the wrong stuff), while simultaneously arguing that the frog is experiencing a significant percept that is the foundation for more complex conceptual abilities. These claims are in tension with each other and he needs to resolve this ambiguity by either dropping his intentional locutions which imply conscious awareness or if he does indeed think that frogs have flies as the content of their experiences he needs to justify this claim not merely assume its truth.

Here I am not saying that Hurtford needs to solve the hard problem of consciousness before proceeding further. But I think he needs to be clearer in what he means when he speaks of frogs having meaningful percepts. If we think of the case of mental imagery and whether it is used in thought (see Kosslyn vs Pylyshyn) this debate is usually handled at the sub-personal level. So know that when people think of mental imagery and rotate some image in their “minds-eye” topographic images are formed in the brain. However, the debates around these topics centre on whether the brain uses these topographic images on the brain in thinking or whether at bottom the brain uses a language of thought for all thinking. The point is that in these debates describing accurately the phenomenology of mental imagery is not considered important. Behavioural tests and neuroscientific studies are the tools typically used. People’s phenomenological reports are treated in different ways by different theorists. They are sometimes treated as mere verbal descriptions of the way things “seem” although no actual images are displayed to a subject (see Dennett 2003). Or the mental images can be treated in a realistic manner as something that a person actually experiences (something with real content), we can explain the neural correlates of these experiences but we cannot explain why these experiences arise from these correlates (See Chalmers). When talking of frogs distinguishing flies from non-flies, or of frogs having meaningful percepts Hurford needs to be clear whether he is speaking at the sub-personal level like Pylyshyn and Kosslyn or if he is speaking at the personal level about frogs experiences.

The debate around on this issue has played a big role in the philosophy of mind and I think Hurtford could have referenced the issue in his foot note. Dan Dennett has discussed this issue on numerous occasions and has argued that there is no fact of the matter as to what the frog intends when he swats at the fly:

“And to the extent that there is nothing in the selective environment that uniquely singles out a particular class of occasions, there is also no fact of the matter about what the frog’s eye report really means.” (Dennett: Intuition Pumps p.257)

“Suppose scientists gather up a small population of frogs from some fly-grabbing species on the brink of extinction, and puts them under protective custody in a new environment-a special frog zoo in which there are no flies, but rather Zoo keepers who periodically arrange to launch little food pellets past the frogs in their care. To the keepers’ delight, the system works; the frogs thrive by zapping their tongues for these pellets, and after a while there is a crowd of descendent frogs who have never seen a fly, only pellets.” (ibid p.258)

Dennett notes that what happens to the frogs in his thought experiment happens all of the time in evolution. It is a case of exaptation where a particular piece of machinery is selected for a different function. To make the case clearer Dennett supposes that in the new environment, variation in pellet detecting ability meant that certain frogs were more likely to survive than other frogs. He further argues that there was no particular moment when we are justified in saying that this is the point where what the frog’s eye report means changes. He argues that there is no fact of the matter about what a frog’s eye report means, and that it is a mistake to think that there is some determinate meaning encoded in the frog’s brain in terms of some kind of mentalese. The meaning of the black dots on the frog’s retina isn’t determined by some central meaner in the brain, rather it emerges gradually through shifts in environmental conditions. He argues that without the “indeterminate” variation in the triggering conditions of the frog’s eyes, selection for a different function would not be possible (ibid p.257).

In Fodor’s (2011) ‘What Darwin got Wrong’ (co-authored with Piattelli-Palmarini) he takes the direct opposite approach to frogs fly snapping and uses it as a tool to beat the concept of natural selection. Fodor discusses Frog’s eating flies interms of the ‘Selection-Selection for distinction’. He outlines Gould and Lewontin’s argument in detail and says that the argument has more serious consequences for adaptationism than they realised. To make this point they discuss the issue of the heart pumping and the heart making thump-thump noises. The heart plays vital role in pumping blood so evolutionary theorists typically argue that the heart was selected-for pumping blood. The problem is that every time the heart pumps blood it also makes thump-thump noises. Now on this issue evolutionary theorists will argue that thump-thump noises are free-riders which piggy back on the selected-for property of pumping blood. Fodor’s problem is that while he assumes that there is a fact of the matter as to whether trait T1 or T2 plays the primary function, he thinks that there is no physical mechanism in the TNS (Theory of Natural Selection) which can distinguish between T1 and T2 so he thinks that there is serious problems with the selection-for story.

Fodor argues that this problem of co-extensive traits appears in behavioural learning theory as well. When we perform conditioned response experiments on Rats we learn that certain responses can be elicited from certain stimulus. However, the problem is that particular responses are massively undetermined by the data of experience. So suppose in an experimental situation we train a rat in stimulus stimulation. Now to do this we reinforce the rat in the presence of a yellow triangle (Stimulus A) but don’t reinforce it in the presence (Stimulus B) a card with an X on it (Fodor and Piattelli p. 103). When the training is complete the rat will produce a particular response when and only when Stimulus A is presented. Fodor notes that in this situation when we ask what the rat has learned in this experiment we don’t know because the Rats behaviour is underdetermined by the data of experience.

Fodor correctly notes that using Mill’s Method of Differences we can whittle away various different possibilities of why the Rat is behaving in the way it is. He mentions learning theory to show how the situation between learning theory and the selection-selection for in the TNS are analogous. However in the case of learning theory we have clever experimenters who can whittle down the options to decide what the fact of the matter is. While in the case of TNS which theorists claim selects for trait 1 as opposed to 2, there is no mother nature to do the clever experiment, and of course in the course of evolutionary history when Hearts were developing there were no clever humans to do the relevant experiment either. We can do experiments now to decide what the fact of the matter is in terms of what is selected for, but mother nature would have been blind to this data. Fodor sums up his argument so far as being that the moral of the story is that to decide what is learned (or selected-for) we need to appeal to counterfactuals (which can be explored in experiments), but “Mother Nature” is blind to counterfactuals so cannot in principle “select-for” one of two co-extensive traits.

Fodor on the Naturalisation of Content:

Fodor argues that the selection-for problem is not limited to learning theory and the TNS but the same problem arises in theories of content. He tries to demonstrate where this problem arises in his theory of content. To do this he discusses the famous question of Frogs swatting Flies. Here is Fodor on Frogs:

“In a nutshell: if the assumption of local coextensivity holds (as of course, it perfectly well might), then fixing the cause of the frog’s snaps doesn’t fix the content of its intention in snapping: either an intention to snap at a fly or an intention to snap at an ABN (ambient black nuisance) would be compatible with a causal account of what the frog has in mind when it snaps. So causal accounts of content encounter a selection-for problem: If something is a fly if and only if it is an ABN, the frogs behaviour is correctly described either as caused by flies or as caused by ABNs. So, it seems, a causal theory of content cannot distinguish snaps that manifest intentions to catch the one from snaps that manifest intentions to catch the other” (Fodor and Piattelli p. 108)

Like in the case of TNS and Learning Theory Fodor argues that the solution is found by appeals to counterfactuals. Fodor goes on to argue that since fly swatting is a behaviour of phenotypes then the problem is one that is raised for the TNS. The TNS is not capable of deciding whether Frogs were selected for Fly swatting or ABN swatting. He reiterates the point again as follows:

“Rather it’s that the individuation of traits depends on the truth of counterfactuals: since (by assumption) every fly-snap in the actual world is an ABN-snap and vice-versa, selection between fly-snappers and ABN-snappers must be sensitive to the counterfactual consideration that ABN-snapping gathers no flies in worlds where the ABNs are BBs, rather than flies. It’s a nice thing about intentional systems that they are sensitive to merely counter factual contingencies. It means that beliefs can take account of what the outcomes of actions would be if…and the believer can act accordingly.” (ibid p.121)

Of course as Fodor emphasizes again and again since selection is not an intentional process it cannot use counterfactuals to decide whether the fly is selected for fly snapping or ABN snapping. So Fodor thinks that this reveals that there is a problem with the TNS, and causal theories of content.

Now the obvious reply to Fodor is that there is no fact of the matter as to whether the Frog is snapping at flies of ABNs. Once we accept this fact we dissolve the problem that Fodor thinks is so perplexing. If the selective environment doesn’t distinguish between ABN’s and Flies then there is no fact of the matter as to whether the fly intends to snap either so there is nothing that TNS is missing out on in this case. Fodor, of course disagrees with the claim that there is no fact of the matter about what frogs intends. He argues that it is plausible that Frogs intend to snap at flies, but says that if you don’t agree on this you can move up the phylogenetic ladder until you are satisfied that the creature intends/believes that x is the case.

Dennett correctly notes that that the problem with this suggestion of Fodor’s is that it assumes that there is a clear fact of the matter for creatures a step up the phylogenetic ladder from frogs either they have concept x or they don’t. However there is little reason to think that this is true. Evolution doesn’t work in the clear cut way that Fodor’s intuitions would like, this is not a black mark about evolution, it is a black mark against Fodor’s pre-theoretic intuitions.

My primary point here isn’t to critique Fodor’s views on evolution (I have already done so in an earlier blog), nor is it to say that Dennett is 100% correct in his views on what a frog’s eye tells its brain. Rather I want to point out that Hurford passes over the topic of Frog percepts too quickly and uses language which is ambiguous as to whether he was using personal or sub-personal levels of explanations. I think he could have at least referenced some of the literature on this topic in a foot note. That said overall his book was one of the better popular explanations of language evolution out there. Now that I have discussed his views on proto-concepts I plan a review of the entire book later.