Fictionalism about Neural Representations

Abstract

1. IntroductionThis paper articulates and explores a novel form of Mental Fictionalism: Fictionalism about the neural representations posited by cognitive science. Cognitive science appears to be committed to neural representations. These representations are claimed to be the springs of our thought and action: they drive our behaviour, determine our thoughts, memories, and inferences. However, despite the central role of neural representations in cognitive science, it is hard to explain what is meant by 'representation' in a way that does not incur problematic commitments. The representations in question clearly cannot be conventional representations that gain their representational content and status through our social conventions; for we are rarely aware that such representations exist, and no adequate social conventions regarding them appear to be in play. The standard reply is that neural representations are representations of a different sort: original or natural representations. This class of representations gain their representational status independently of, and in some sense prior to, our social conventions. But what is a natural representation? Attempts to answer this question-naturalising representation-have been on-going since the 1970s. Unfortunately, this project to date has been largely unsuccessful. Many contemporary theorists are sceptical that an adequate naturalistic theory of representation will ever emerge.For this reason, some theorists have been drawn to Eliminativism about talk of neural representations in cognitive science.1 If cognitive science could stop appealing to neural representations, then there would be no need to give an account of representation, and therefore no need to give a naturalistic account. However, thoroughgoing Eliminativism about neural-representation talk is a hard road to follow. Although some cognitive phenomena can be explained in nonrepresentational terms, other aspects of cognition appear stubbornly resistant to nonrepresentational explanation. Attributing neural representations seems to be the best way to explain many of our cognitive abilities.2Cognitive science appears to face a dilemma: either it uses neuralrepresentation talk and is lumbered with the task of naturalising representation, or a radical and undesirable revision to the practice of cognitive science is required.Neural Representation Fictionalism (NRF) offers a neat way out. NRF opens up a third option: allowing us to use neural-representation talk in cognitive science without the cost of naturalising representation. NRF promises to rid us of one of the biggest problems facing representation talk in cognitive science without the pain required by Eliminativism. NRF purports to deliver the benefits of both Realism and Eliminativism with the costs of neither. The only downside of NRF is that it would require us to reinterpret neural-representation talk in cognitive science in a fictionalist way. At least on the face of it, it is not obvious that this is not a price worth paying. NRF seems worth exploring.Fictionalism about a given discourse is the view that claims C in that discourse involve genuine statements of fact-they aim to describe the world-but, in contrast to Realism, those claims C do not aim at truth. Instead, they serve some other purpose. A fictionalist might remain agnostic about the truth value of her claims C (as van Fraassen [1980] does), or she may declare that the claims C are literally false in spite of their cognitive value (as do Nolan, Restall, and West [2005]).Fictionalism of many stripes has become popular in recent years. Forms of Fictionalism have been developed for mathematical discourse, moral discourse, modal discourse, and negative existential talk. In each case, the motivation bears a striking resemblance to the problem facing cognitive science above. We have a practice-mathematical talk, moral talk, modal talk, or negative existential talk-that appears to commit us to the existence of troublesome entities-numbers, moral facts, possible worlds, nonexistent objects. …