Inference to Representation: Scientific Explanation & Two Kinds of Eliminativism

The first EI of term (Feb 12th, 12:00-12:50, Pevensey 2A11) will be led by Jonny Lee:

Inference to Representation: Scientific Explanation & Two Kinds of Eliminativism

Representation features heavily in scientific explanations of cognition. The principle of ‘inference to the best explanation’ (IBE) says that we ought to believe that our best theories are true. If our best theories feature representations then, according to IBE, we ought to believe that those ascriptions are true. At the same time, eliminativism about mental representation remains popular. In its various forms, eliminativism holds that we ought to reject talk of representation in some or all of cognitive science. How do we reconcile the fundamental role of representation in cognitive science with the continued appeal of eliminativism? In beginning to answer this question, we must understand the different motivations belonging to distinctive kinds of eliminativism. There is more than one kind of eliminativism, and the conflation of these kinds leads to conceptual confusion. I will argue that there is an important distinction between a priori and a posteori eliminativism, and furthermore that a priori eliminativism faces serious obstacles in establishing itself as a persuasive challenger to the status quo. I will finish by considering the possibility that the ontology of mental representation could turn out to be more nuanced than philosophers have previously allowed for.

Audio (10mb, .mp3)

7 thoughts on “Inference to Representation: Scientific Explanation & Two Kinds of Eliminativism

  1. Jonny,

    Thanks for the interesting, clear, well-argued talk on a very “EI” topic. A great start to the term. I have a few specific comments. Some of them I already voiced at the talk, but I am including them here for the sake of completeness.

    As I mentioned to you briefly after the talk, I saw you as making two points that were relatively independent of each other:

    – Fictionalism (Fictivism?) is an alternative to the realist and eliminatist positions which have dominated the debate on representations;
    – One’s metaphysics of representation need not be monolithic: One can question the reality of one kind of representation while embracing the reality of another.

    I like both of these. You didn’t have time to do this, but I personally would like to hear more about the first point:

    – Is there anything of substance that hinges on whether one is a methodological (fictive) representationalist vs a realist representationalist? I don’t recall what you said when I raised this before.
    – But perhaps more central to your talk, isn’t the fictivist rejecting IBE? To embrace representationalist talk seems tantamount to saying representationalism is the best explanation, so if one also withholds endorsement of its truth, one is rejecting the IBE principle, right? Apologies if you already addressed this.

    Concerning the second point: isn’t the case for this already made by Ramsey when he introduces the S-representation/”receptor representation” distinction? Or is there some way you are going beyond what Ramsey says? In particular, I think you need to make it clear that this position is not tenable if one is an a priori eliminativist (assuming you agree!).

    Also, when you were characterising a priori eliminativism, it sounded like you said that it involved the conclusion that our best theories of content are false (or did I mishear you?). If so, that would mean that a priori eliminativists are rejecting IBE. Is that what you take them to be doing?

    • Thanks Ron! I’ll try to sketch out a reply to your questions and concerns.

      “Is there anything of substance that hinges on whether one is a methodological (fictive) representationalist vs a realist representationalist? I don’t recall what you said when I raised this before.”

      In short, no, I don’t think so. Which is to say, for any given ascription of representation, it’s not obvious what difference it would be make to “merely treat” it as a true ascription, versus treating it as a true ascription. Put another way, if one individuates representations the same way, regardless of whether one thinks of those representations as fictional or real, then there doesn’t seem to be anything interestingly different, at least for the purposes of cognitive science. This is a good thing for the fictionalist! They (presumably) want their position to be compatible with the use of representation within cognitive science, whilst conveniently avoiding the need to provide a naturalised theory of representation.

      Having said this, it’s important to, firstly, keep these positions apart if we’re interested in the metaphysics of representation, secondly, be aware of the variations in motivation for these umbrella positions. The precise motivations for one’s realism or fictionalism will determine how you individuate the ‘real’ or ‘fictional’ representations. Within those positions, there will be disagreements over how best to carve up representations (even if, for any one ascription of representation, your ontological position makes no difference). Perhaps you’re a fictionalist, but maintain that only certain domains of cognitive science should preserve representation talk, because it is only within those domains that representation talk has its epistemic value. Alternatively, you might be a realist, but have high standards of what counts as a genuine representation, in which case your representations are going to be different from a liberal fictionalist who is happy for representation talk to remain more or less unchanged from the historical practices of cognitive science.

      “…isn’t the fictivist rejecting IBE? To embrace representationalist talk seems tantamount to saying representationalism is the best explanation, so if one also withholds endorsement of its truth, one is rejecting the IBE principle, right?”

      This seems like the right worry to me! It appears difficult to maintain IBE and be a fictionalist at the same time. I mentioned this in passing as the ‘challenge from IBE’ (as also faced by a priori eliminativism). Obviously, this is a problem if you think that getting rid of IBE is too high a price to pay. Mark Sprevak discusses this in a 2013 paper titled ‘Fictionalism about Neural Representations’.

      I think as a fictionalist, you could think that that our best explanations never invoke representations really, but that treating some aspect of those explanations as representations serve some ineliminable epistemic purpose; our best explanations actually aren’t representational at all, but nevertheless we’d best keep treating them as representations. Or you might think that some of our best explanations invoke real representations, but some invoke representations that, epistemic needs being what they are, must be treated as fictions (given your standards of what counts as a real representation), hence the possibility of a more ‘nuanced ontology’. A lot here hinges on whether fictionalism is compatible with IBE.

      Here’s one thought: akin to the a priori eliminativist response to the challenge from IBE (as I briefly mentioned in my talk, also see below), the fictionalist could hold that the concept of representation contains something explanatory, but its contentful properties preclude representation from ever being fully realised at the sub-personal level. In other words, part of what’s covered in the concept of ‘representation’ is explanatory and found at the sub-personal level, but not all the conditions required for complete representation are met. Unlike the eliminativist though, the fictionalist could hold that we can’t get rid of the bad whilst holding onto to the good, thus we cannot get rid of representation talk altogether. (Take this idea with a pinch of salt though, I haven’t considered if and how this could work).

      I didn’t want to go into too much detail about this problem, firstly, because I think the problems for the fictionalist are basically the same as the problems of the a priori eliminativist that I spent a little more time focussing on. Secondly, because for the purposes of the talk, I wanted to put these problems aside and highlight the possibility of compatabilism between realism, fictionalism, and some kind of eliminativism, across different domains of cognitive science. This might have been a bit hasty, as it might hard to see how this could even be possible without a more fully realised account of fictionalism.

      I believe Adrian Downey has considered how the fictionalist about mental representation can hold onto IBE.

      “…when you were characterising a priori eliminativism, it sounded like you said that it involved the conclusion that our best theories of content are false (or did I mishear you?). If so, that would mean that a priori eliminativists are rejecting IBE.”

      In short, the a priori eliminativist can either reject IBE, or maintain IBE and offer an account of why our best theories do not (really) ascribe representations.

      What I intended to say in the talk was this: for the a priori eliminativist, ascriptions of representations are false, and they are false in principle because they make a category mistake. I assumed that the a priori eliminativist did adopt IBE, and held that the claim, “our best theories imply the true ascriptions of mental representations”, is false. A natural consequence of this position is the thought that if you think our best theory of a phenomenon implies ascriptions of representations, you’re wrong about what the best theory is.

      Later, when I was attempting to sketch an a priori eliminativist response to the ‘challenge from IBE’ (the challenge of responding to the apparent ubiquity of representation in what appear to be our best explanations), I mentioned the possibility that the a priori eliminativist could accept that our best theories do actually invoke representation. Under this account, ascriptions of representations are doing something productive, not because of representations per se, but because there is something contained within the concept of representation which is explanatory (for example, perhaps the concept ‘representation’ captures some property of tracking distal objects. You might think of representation as historically the closest folk notion to hand for capturing something important about an explanandum, but one that ultimately brings too much baggage with it. Unfortunately, by importing that folk notion we’ve also imported the problematic notion of content. We should therefore keep whatever is good about representation, whilst exorcising content.

      I mentioned at the beginning of the talk that one could reject IBE altogether, in which case the burdens and concerns of the debate over the ontology of mental representations in cognitive science will shift.

  2. I’m sorry too, Jonny and Ron, to have missed the re-inaugural of E-I. It would not have been impossible to get there but I did fear the treatment would be a bit to technical for my priorities. Ron’s comments suggest not.
    My position is ‘simply’ old-fashioned verbal hygeine: people / whole systems re-present aspects of reality in words, pictures etc. – observable ‘symbols’ of what they represent. No physical, social or mental part of the system doing the representing can literally represent. We need to be very cautious with metonymy.
    Does “fictive” as a middle way risk taking tendentious jargon too seriously in ontology? I have no doubt that patterns at receptors (and at motoneuron pools) contain information that contributes to the content of what the individual is doing, saying etc.: such ecology is central to my calculations of how a mind is working. I don’t see why we need or even want to call part of the content of a message what the individual represents by the whole transfer of information.
    – David Booth

    • Hi David. Sorry you couldn’t make it and thanks for your thoughts.

      I think your position sounds like a sensible one. Though I don’t know the details of your thoughts on the matter, I’d say it’s a position I’m broadly sympathetic to. However, I do think there might be good reasons to think of ‘parts of systems’ as representations (though I must admit, whilst I understand the ‘persons vs. non-persons distinction, I’m not sure I have a good grip on ‘system’s vs. system parts’, as clearly marking anything ontologically serious).

      In any case, and without trying to open a whole can of worms, here’s what I think: cognitive scientists do (sometimes) identify sub-personal mechanisms with a distinctively representational function as part of a larger system, a function which can be given a fairly robust definition which meets the “job description” of representations in the ordinary sense of the word. You might think of these mechanisms as mere “as if” representations, but I think there are good grounds for not distinguishing between mechanisms which play a very representational-like role, and the “really real” representations. I personally think a lot of this is helped by prising apart the issue of what it is for a mental state to function as a representation, and the question of content, what a representation is about. (See Ramsey, 2015, “Untangling Two Questions about Mental Representation”).

      In any case, I agree with the importance of ‘verbal hygiene’. I think it’s important not to stretch our notions of representation and content to the point of terminological failure. Seriously characterising something as a representation must involve looking at what explanatory work (qua something recognisably representational) that characterisation is doing. Perhaps I’m just a little more hopeful about the prospects of a strong account of representation that respects certain practices in cognitive science, and our everyday notion of representation.

  3. Having just listened to the seminar online, I’d like to pitch in with some comments.

    Firstly, I see explanations as instrumental tools for helping us think about the phenomena they purport to explain. Hence, the “best” explanation is only best relative to an individual’s current explanatory purposes and cognitive landscape.

    Almost all representationalists also believe that the phenomena they are interested in can also be accounted for in purely physical terms, i.e. by a hypothetical account that does not invoke representations. In this sense, the typical cognitive scientist believes that “underlying” any representationalist account of physically observable phenomena, there is a non-representationalist physical account.

    Neither of these explanations is universally “better” than the other; they have different instrumental properties that make them suitable for different endeavours.

    I am far less sympathetic to the notion of “verbal hygiene” than Jonny; it’s commonplace for scientific disciplines to use everyday words in specific technical senses. Moreover, technical scientific concepts can sometimes drive changes in everyday usage and everyday conceptual topography – for instance, the word “fish” is no longer used to refer to whales.

    For reasons of brevity, I will restrict my initial post to these comments, although I have lots more!

    Warm regards,


  4. Simon, we’re having a detailed conversation about representations within systems on (and off) the email list of the Embodied Cognition Reading Group. That email exchange can be ‘archived’ on PAICS, Ron says, and we can continue there if you wish (and others can join in – including Jonny if he wants to go into more detail).
    Here I’m just adding a comment on your response to my plea for ‘verbal hygeine.’ My position comes from bitter experience in behavioural neuroscience and purportedly biological psychology. Psychology in general should be in the business of “using everyday words in specific technical senses” when they are about what people do, think and feel. What ordinary language should not be used for is to confuse, and even to dodge difficulties of, the conceptualising and testing for the mental, material and societal mechanisms by which people act, perceive, reason and ’emote’. – David

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s