Neuroscientist John Kubie criticised a recent claim I made in my Blog-Post ‘What is a Concept and What Creatures Have them’. He argued as follows:

 “Your second sentence “A concept is to some degree a theoretical term that we use to explain, predict and control the behaviour of various different creatures” seems wrong to me. For me the core sense of “concept” has nothing to do with communication or control of other creatures. It has to do with aggregation of diverse stuff into a single “symbolic?” Entity. It’s a logical operation.”

I think what I wrote in this instance was a bit ambiguous so it may need some clarification. I was speaking about the evidence we use to justify our attribution of a concept to another creature and noting the importance of behavioural evidence in such an attribution. So, for example, consider the paradigm where researchers try to discover what conceptual abilities pre-linguistic children have through monitoring their surprise reaction. A child is typically habituated to a particular phenomena, e.g. a ball passing behind a screen and coming out the other side. However, in experimental settings after the child is habituated to the particular phenomena they are presented with strange behaviour of the ball. So one ball may pass through a screen and two balls will come out the other side. Or a ball may go behind the screen and not come out the other side. If a child looks surprised at these events we will assume that they have some expectations of how objects should behave and are surprised that the objects do not typically behave in this way. Now Piaget and Quine both criticised this experimental procedure claiming that it merely shows that children have expectations of continuity in their environment not that they have concepts of Object, Number etc. And in recent times Carey, Soja have made the experiments more sophisticated to avoid the objections of Piaget et al. Now my point here isn’t to evaluate the theories of Carey et al but it is merely to indicate that in these experimental settings Concepts are attributed to children because of the explanatory role they play in the experimental theories of theorists. Or outside of the lab people may attribute a concept a fly to a frog because of its behaviour around flies (in this instance they would probably be wrong). In these settings we use our attributions of concepts as part of an explanatory schema to explain what the creature will do, and will modify our attributions if the creatures behave in ways that we did not expect.

When John argued that:

 “The core sense of “concept” has nothing to do with communication or control of other creatures. It has to do with aggregation of diverse stuff into a single “symbolic?” Entity. It’s a logical operation”

I am confused as to what he meant in the above claim. Is he offering a theoretical definition of the term ‘Concept’ which he thinks will make our theorising easier?  Is he in a sense offering a convention as to how the word ‘Concept’ is to be used in scientific theorising? Or is he arguing that his definition corresponds with the ordinary language use of the word ‘Concept’? I would argue following the philosophical work of Wittgenstein and the scientific research of Lakoff and Rosch that there is no one accepted meaning of ‘concept’. Rather the word concept is a family resemblance term (though people may have a prototype of concept in mind then they use the term).  Either way I would need to know what John has in mind when he argues that a concept is “an aggregation of diverse stuffs into a single symbolic entity” before I can adequately reply to his comment.

In my blog ‘What is a Concept and Which Creatures Have Them’ I criticised philosophers who think that only language using creatures have concepts. I argued that these theorists are acting like Platonists about the nature of concepts and ignoring recent empirical research into the subject. Philosopher Jim Hamlyn has long argued that animals have no concepts unless they are language users and tool users.   So, for example, while Dolphins may be one of the most social and intelligent animals on the planet because they don’t use tools he would argue we are not justified in attributing conceptual abilities to them. In his recent very interesting blog ‘Tools in the Work Shop of Language’ Jim defended and clarified his position on animal concepts.

One thing very clear from the blog is that Jim adopts a guilty until proven innocent approach to attributing concepts to non-linguistic animals. He argues as follows:

“Secondly, the assumption that certain sophisticated behaviours can only be explained by concept-possession is only justified if every other explanatory alternative has been ruled out.”  
This seems a very strange criterion to offer as when a theorist is justified in attributing a concept to another creature. A theorist typically doesn’t say that a behaviour can only be explained by concept attribution; rather they would argue that attributing conceptual abilities to these creatures is the best way of explaining their behaviour.

By setting things up in this way Jim gives himself an easy task.  He argues that people who attribute concepts to non-human animals claim that ONLY conceptual explanations are possible, so If Jim can tell any story that doesn’t attribute conceptual abilities to non-human animals he has automatically refuted his opponent. Except of course he hasn’t refuted his rival theorist because the rival theorist claims that the conceptual explanation is the best explanation not that it is the only one.

Jim seems to start with the assumption that we should assume from the outset that favouring the conjecture that animals have no concepts is favourable to the view that they have concepts. Here is an example of this:

“Experience leads to the development of expectations about the regularities of the universe and of unfolding events. I propose that we rule out all possible nonverbal explanations before we are tempted to ascribe capacities of conceptualisation to nonverbal creatures.” (Hamlyn ‘Tools in the Workshop of Language’)

By adopting this question begging approach he  shifts the burden of proof onto his opponents. This approach is though problematic, it is not always easy to decide whether view A or view B should be accepted as our null hypothesis. In her recent book Kristin Andrews spelled out the difficulties in deciding what our null hypothesis should be:

“But perhaps a greater problem arises at the point of deciding on the null hypothesis itself, because beginning an investigation of a property with a sceptical view may introduce against animals having that property. When our concerns are purely epistemic, as they presumably are in the case of animal cognition, it isn’t clear why either the sceptical or optimistic hypothesis should get preferential treatment from the outset” (Andrews: ‘The Animal Mind’ p. 42)

I think Andrews has a point; in debates about whether we should attribute various mental abilities to animals or not people do seem to operate with the null hypothesis that it is simpler to avoid attributing things like concepts to animals if we have an alternative story. Marion Stamp Dawkins in her ‘Why Animals Matter’ adopts this approach.  Andrews notes that one there are some good reasons to not adopt the null hypothesis in the way people like Dawkins does. She cites Eliot Sober (2012) and Frans de Waal (1991) have argued that given facts about convergent evolution, and continuity of life on earth the opposite assumption should be made. Sober’s idea is that if humans and closely related species (Apes) exhibit both exhibit behaviour X and humans produce behaviour X by occupying mental state m then we have evidence that M is the proximate mechanism that the chimpanzees employ in producing X (ibid p.43)[1]

I personally prefer to adopt a wait and see approach; I think Sober’s approach is ambiguous as to how far down the species we can go and assume mental continuity. While those who prefer to make the opposite assumption typically do so by arguing that their approach is simpler but they typically never specify precisely how they are judging simplicity. In my view if we have two theories one of which says that Dogs are concept users, and the other which says they are not, if we do not have sufficient evidence to decide the issue we should continue our search for the correct answer; not make question begging assumptions as to which approach we should assume is true by default.

As his blog-post goes on Jim argues that Jerry Fodor’s language of thought thesis is impeding progress into the study of animal mental capacities. This is news to me. Fodor’s influence in cognitive science is pretty much a whipping boy. He is one of the most cited theorists in the philosophy of mind but is typically only cited so people can say why they disagree with him. Researchers in to the conceptual abilities of pre-linguistic children such as Spelke, Carey, Soja, Markman etc have all been very critical of Fodor’s views on concepts. Susan Carey in her (2011) book ‘The Origin of Concepts, is very critical of Fodor’s views on Concepts. I rarely see his name mentioned in the animal ethnology literature I think to focus on Fodor’s views on animal conceptual abilities in the context of this discussion is radically wrongheaded given how most cognitive scientists disagree with him.

When dealing with proponents of a particular position I think it is best to focus on the best theorists holding a particular position. In defence of his position Jim mentions the arguments of Davidson as being very compelling against the position that animals have concepts. In an earlier blog evaluated Davidson’s arguments in detail and critiqued them all. I think Jim would not find it very helpful if I pretended that Descartes (who held that animals were unthinking machines) represented the best current evidence in favour of Jim’s position. Likewise I respectfully submit that Jerry Fodor doesn’t represent main stream cognitive science and if Jim wants to criticise representational views in Cognitive Science that refuting Jerry Fodor is not the way to go.

Jim ends his blog-post with an interesting discussion of what he takes to be a problem with attributing concepts to animals.  He notes correctly that an animal learning a few proper nouns does not mean that the animal can manipulate them and combine them in the way language users do. He argues that a creature who has concepts but who doesn’t have the skills to use them in productive ways has nothing more useful than a stone, or a stick that they don’t know how to use.

I think Jim’s emphasis on the analogy between concepts and tools is a good one and warrants further exploration. When we attribute concepts to animals we are arguing that they can use these concepts to think and decide what to do. When making this claim we need to be very particular about what we are claiming. Any claim than an animal can use concepts in some internal language of thought by combining them in certain ways needs to be very clear in the behavioural evidence for how these concepts can be combined and cannot be combined. This type of evidence is rarely if ever presented and until it is I think sceptics about animal concepts will have a point. In his (2012) paper ‘Why we cannot say that Animal’s Think’ Joseph Beck argued that we are not justified in assuming that animals have concepts because we have no theory of how they systematically use the concepts. Beck followed Evans (1982) in arguing that if we are to attribute conceptual abilities to a creature the creature must be capable of using these concepts systematically. Beck was responding to Susan Carey[2] work which attributed concepts of numbers to some animals. By analysing the experimental evidence of Pigeons behavioural responses he demonstrated that the Pigeons were incapable of using these concepts in a way that represented systematicity and generality constraints. So he concluded that we are unjustified in assuming that Pigeons are thinking using concepts. I think that this constraint is a good one to help refine research into the area (Andrew’s claims that Caruthers (2009) demonstrated that animals do satisfy a weak version of the generality constraint), those of us who claim that non-linguistic animals have concepts need to meet this criterion. So this leaves a lot of empirical leg room in the debate and I think Jim has a point when he argues that the ability to use a concept for various purposes (which requires some kind of systematicity), is the key area to be researched.

[1] Andrews got the quote from Sober (2012) pp 3-4

[2] Here I am following Kristin Andrew’s explication of Beck’s work in Andrews (2015)