Two systems accounts of cognition and the nature of judgement
30–31 March 2016
Workshop with Till Vierkant (Edinburgh), Suilin Lavelle (Edinburgh), Josefa Toribio (Barcelona), Stephen Butterfill (Warwick), Josh Shepherd (Oxford), and John Michael (Budapest)
Venue: Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Luisenstraße 56, 10117 Berlin, Room 144 (ground floor)
Hosts: Richard Moore and Till Vierkant
This workshop will bring together philosophers working on two system approaches to social cognition and philosophers working on the nature of judgement. In social cognition there has been a debate about the nature of early belief attribution processes and the form that System 1 processes would need to take for early belief attribution to happen there. At the same time it has been argued that judgments which seem quintessentially System 2 cannot be System 2 processes at all given standard definitions. Both debates grapple with achieving a good understanding of the nature of the two systems beyond the usual dichotomies. Bringing together these two groups will cross-fertilise both debates and help to achieve a clearer picture of what it means for a process to be either System 1 or System 2.
30 March 2016
12:30 Coffee and fruit
13:00 Till Vierkant, S2: Judgment Day
14:30 Josh Shepherd, Control, Decision, Judgment
15:50 Coffee break
16:20 Josefa Toribio, The Judging Skill
31 March 2016
9:00 Coffee and fruit
9:30 Stephen Butterfill, How to Distinguish Two (Or More) Systems for Social Cognition
11:00 John Michael, If there is a system 1 for belief reasoning, then it is a rather system 2-ish system 1
14:00 Suilin Lavelle, Beliefs all the way down? Challenges to 'One System' approaches to mindreading
15:30 Closing discussions
Suilin Lavelle: Beliefs all the way down? Challenges to 'One System' approaches to mindreading
It is uncontroversial that a significant portion of the cognitive processing facilitating social interactions happens quickly and efficiently. There is, however, considerable dispute concerning the nature of the psychological concepts that can be processed quickly and efficiently. On one side of the debate are those philosophers and psychologists who maintain that such psychological concepts cannot be representational, because then they cannot be processed quickly and efficiently (Apperly and Butterfill 2009, 2013). On the other side are those who maintain that representational states can be processed quickly and efficiently (Carruthers, 2011, 2013, 2015). These are the two-conceptual system and one-conceptual system views, respectively.
This paper challenges the claim, made by one-conceptual system views, that fast efficient mindreading is simply mindreading which does not involve working memory and/or executive function. It does so first by questioning the nature of the representational states that one-system accounts maintain can be processed quickly and efficiently, claiming that once these states have been pared down such that they can be processed quickly, they end up looking quite different from our folk psychological concepts. This pushes towards a two-conceptual systems approach. Second, the one-system account is committed to working memory and executive function being predictors of success in certain mindreading tasks. I maintain this connection is not supported by empirical data, and that the terms ‘working memory’ and ‘executive function’, in this context, are used too loosely and mask a variety of more nuanced phenomena.
Till Vierkant: S2: Judgment Day
According to the very influential two system accounts of the mind there are two distinct forms of processing in the human mind. In such accounts system 2 is supposed to contain a specifically human form of judgment and decision making. This paper wants to demonstrate that on two very plausible premises it is impossible that system 2 can contain any judgements. The premises in question are the claim that system 2 processing is always intentional and that doxastic involuntarism is correct. The paper examines the premises and discusses a number of potential objections to the argument. None of these objections proves decisive and as the premises are very plausible the argument seems sound. The paper finally considers three potential ways of dealing with the consequences of the argument.
Josefa Toribio: The Judging Skill
In this talk I argue on two fronts. First, I press for the view that judging is a type of mental action, as opposed to those who think that judging is involuntary and hence not an action. Second, I argue that judging is specifically a type of non-voluntary mental action. My account of the non-voluntary nature of the mental act of judging differs, however, from standard non-voluntarist views, according to which ‘non-voluntary’ just means regulated by epistemic reasons. In addition, judging is non-voluntary, I contend, because it is partially constituted by the exercise of a non-reason-governed skill. This skill, which I call ‘critical pop-out’, consists of an unreflective, often unconscious, ability to detect the kind of situations in which the reflective abilities that also partially constitute our acts of judging should be deployed. We are responsible for our judgments, I conclude, because in identifying such reflection-inviting situations, we reveal the kind of epistemic agents we are.
Joshua Shepherd: Control, Decision, Judgment
In earlier work I have defended the view that decisions -- momentary mental events of intention formation -- ought to be considered intentional actions. Till Vierkant has criticised this view. Further, he has noted that the arguments I offer can be applied to a view on judgments, and he has criticised such an extension as well. In this talk, I: 1] lay out my argument for the view on decisions, and consider Vierkant's criticism of it, 2] consider whether the argument can be extended to judgments, and 3] consider Vierkant's criticism of such an argument regarding judgment. My hope is that in doing so, the motivation for and functional significance of my earlier argument, and the nature of my disagreement with Vierkant, will be brought into sharper relief.
Stephen Butterfill: How to Distinguish Two (Or More) Systems for Social Cognition
Suppose you observe two exercises of competence in social cognition and want to know whether the systems underpinning the exercises are the same or distinct. How can you tell? An immediate obstacle is theoretical: there are many attempts to characterise ‘system’, each only slightly different from the next, and no obvious principle for deciding between competing characterisations. You therefore cannot rely on any one theory. Despite this obstacle, it seems to be possible to formulate hypotheses about the distinctness (or identity) of systems which generate testable predictions. But how? This talk aims to extract a recipe for distinguishing (or identifying) systems from research on understanding speech, action, emotion and belief.
John Michael: If there is a system 1 for belief reasoning, then it is a rather system 2-ish system 1
The two systems theory developed by Apperly and Butterfill (2009; Butterfill & Apperly, 2013) is an influential approach to explaining the success of infants and young children on implicit false belief tasks. There is extensive empirical and theoretical work examining many aspects of this theory, but little attention has been paid to the way in which it characterizes goal attribution. I will argue here that this aspect of the theory is inadequate. Butterfill and Apperly’s characterization of goal attribution is designed to show how goals could be ascribed by infants without representing them as related to other psychological states, and the minimal mindreading system is supposed to operate without employing flexible semantic-executive cognitive processes. But research on infant goal attribution reveals that infants exhibit a high degree of situational awareness that is strongly suggestive of flexible semantic-executive cognitive processing, and infants appear moreover to be sensitive to interrelations between goals, preferences and beliefs. Further, close attention to the structure of implicit mindreading tasks – for which the theory was specifically designed – indicates that flexible goal attribution is required in order to succeed. I will consider ways in which the theory could be revised to address this challenge.