Matt Maguire (ANU, Australia): Restricting Associative Explanations in Comparative Psychology

According to standard practice in comparative psychology, if a researcher aims to prove that nonhuman animals (hereafter, animals) possess some complex cognitive capacity, they must first rule out simpler explanations. This manifests in the following way: a researcher claims that successfully performing the experimental task demonstrates that the target species possesses a complex cognitive capacity (e.g. Theory of Mind, causal reasoning, etc.), a respondent then offers an associative explanation for how the animals could perform the task; if the respondent is successful, we conclude it is likely the animals used the simpler, associative method. In recent years, purportedly associative explanations of thought have rapidly increased in explanatory power, making it increasingly difficult for comparative psychologists to prove that animals possess complex cognitive capacities. This raises concerns that best scientific practice may exaggerate differences between human and animal minds. To avoid this, we can either deny that this standard practice in comparative psychology is best scientific practice (attempts at this have been contentious) or restrict which kinds of associative explanations are: a) genuinely associative, and/or b) genuine explanations of the target behaviour, and/or c) incompatible with non-associative explanations.  

The most prominent attempted solutions to this problem simultaneously pursue options b) and c) by claiming that some purportedly associative explanations do not genuinely explain the target behaviour and are therefore compatible with alternative explanations. For instance, Cameron Buckner divides associative accounts into two categories: associationex and associationmod[1]. In associativeex accounts, associations are genuine mental objects which cause the explanandum behaviour. In contrast, associationmod accounts make no claims about the explanans of the behaviour, but merely state that the behaviour is well-modelled by a link and node structure. The upshot of this is that associativemod accounts are compatible with nonassociative explanations, so do not defeat claims that animals possess complex cognitive capacities. However, this is not immediately useful for the comparative psychologist because Buckner is silent on how to determine whether specific associative explanations are associativeex or associativemod. 

In this paper, I identify which features distinguish associativeex from associativemod accounts. Drawing on the mental causation literature, I argue that associationsex possess a natural cluster of functional features. In my view, there are two functional features which define associationsex: firstly, they are directed causal links between mental representations (i.e. if x is associated with y, then x occurring in the mind will cause y to occur – but not necessarily vice versa); and secondly, they track and are formed from regularities in the animal’s experience. If a purportedly associative account lacks one of these features, then it is likely associativemod. I conclude by sketching a field guide for distinguishing between associativeex and associativemod accounts. If adopted, this would make it easier for comparative psychologists to prove animals possess complex cognitive capacities because many simpler, associative ‘explanations’ could be dismissed as associativemod.  

[1] Buckner, Cameron. ‘Two Approaches to the Distinction between Cognition and “Mere Association”’. International Journal of Comparative Psychology, vol. 24, no. 4, 2011, pp. 314348.