The Narrative Practice Hypothesis: clarifications and implications

Abstract

The Narrative Practice Hypothesis (NPH) is a recently conceived, late entrant into the contest of trying to understand the basis of our mature folk psychological abilities, those involving our capacity to explain ourselves and comprehend others in terms of reasons. This paper aims to clarify its content, importance and scientific plausibility by: distinguishing its conceptual features from those of its rivals, articulating its philosophical significance, and commenting on its empirical prospects. I begin by clarifying the NPH's target explanandum and the challenge it presents to theory theory (TT), simulation theory (ST) and hybrid combinations of these theories. The NPH competes with them directly for the same explanatory space insofar as these theories purport to explain the core structural basis of our folk psychological (FP)-competence (those of the sort famously but not exclusively deployed in acts of third-personal mindreading).

Keywords:

• folk psychology
• reason explanations
• theory of mind
• theory theory
• simulation theory
• narrative

Acknowledgements

I wish to express particular thanks to Leon de Bruin, a visiting researcher from Leiden, who discussed the penultimate draft of this paper with me intensely during his time at Hertfordshire. Equally, I am grateful to Joanna Gillies for her marvellous enthusiasm and her kind (super-obligatory) assistance in looking over and correcting the final drafts of my work and also to Laura Sparaci (a visiting researcher from Parma) who also provided very helpful comments on the final draft.

Notes

The commitment to the existence of such principles is vital to any interesting variant of TT. Crane makes this indelibly clear: ‘For there to be a theory of mental states, then, there needs to be a collection of principles which explain mental phenomena’ (Crane 1995, 63, see also Churchland 1981; Fodor 1987; Botterill 1996; Botterill and Carruthers 1999; Sehon 2005; Borg 2007).

Thus Currie observes that: ‘We simulationists are an anti-intellectual crowd, always keen to minimize the role of theory and to promote the claims of informationally-impoverished strategies like empathy’ (Currie 2004, 191).

Goldman characterizes ‘enactment’ or E-imagination as a form of simulative activity that involves the production and use of mental counterparts (see Goldman 2006, 149; and also Currie and Ravenscroft 2003 for a detailed articulation of this idea).

Theory is meant to do other jobs too. It allegedly generates hypotheses in the form of predictions which are, rather incredibly, then supposed to be tested by running further simulative procedures (Goldman 2006, 184–5). I say ‘rather incredibly’ for I assume that the point of giving FP explanations is to say why someone acted and not to say why someone might have acted. If so, I do not see how this procedure could serve as an adequate test for mindreading accuracy at all. I have discussed precisely this worry about the fitness for purpose of simulation and theorizing when it comes to explaining action elsewhere (see Hutto 2004).

Co-cognizing is essentially the replication of a target's thoughts in oneself, for the purpose of predicting them. This is done by observing the outcomes of one's own thinking (cf. Heal 1998b, 491). In this process we, ‘harness our own cognitive apparatus and make it work in parallel with that of the other’ (Heal 1998a, 85). Our conclusions about what another is thinking can be justified either by appeal to analogy or by assuming that they have a minimal competence in dealing with aspects of a common world.

Stueber defends the view that the best version of folk psychology recognizes its essentially engaged, simulative nature. It is not to be understood as purely detached and theoretical. This is because simulative aspects (understood as invoking capacities for re-enactive empathy), although limited in certain respects, are required. They are allegedly epistemically non-negligible, constituting the central default mode of understanding others in terms of reasons. This is the main claim of his book (Stueber 2006, 19, see also 105, 121, 125). Therefore despite his marking out an utterly central role for theory, Stueber claims its use cannot infect, make superfluous or otherwise obviate the crucial simulative activity that is necessary for understanding minds.

Not everyone agrees about this (e.g., see Ratcliffe 2007, 2008; Hutto 2008c).

I supply detailed reasons for rejecting both theory- and simulation-based theories of purported internal mechanisms that might explain the basis of our FP-competence in full elsewhere (see Hutto 2008a, chaps. 8, 9). My ultimate reason for being suspicious of these positions is that I reject the idea that ‘rules are stored somewhere in our mind/brain … [a conception that] is the groundwork of the modern discipline of cognitive science’ (McGinn 2002, 48). But, independent of this, the existing offerings fail because none of them can adequately explain how we ‘acquire’ the concept of belief in a way that makes sense of the role it must play in such theories (either informing the ToM principles (TT) or coming into play in the attribution phases of mindreading (TT and ST)).

FP-competence equates to the practical knowledge that training by means of FP narratives inculcate. It is this competence that we acquire when we develop the ability to sensitively deploy FP skills. This answers Currie's (2008) question about what the ‘it’ refers to when I talk of knowing how and when to use FP. Any confusion on this score that may make it appear as if I am somehow committed to the existence of FP as a body of knowledge, or set of principles to be wielded, is merely a trick of grammar. Focusing on my other claims about the exercise of this underlying competence also addresses Currie's concern that reasons explanations do not obviously appear to be narratives. For example, as he notes saying that I went out ‘to get milk’ in response to a query does not seem to count as giving a narrative (see Currie 2008). Here appearances are deceiving. Because reason explanations function as normalizing, contrastive explanations to give one's reason in response to a question is, for the well trained, only to tell the most relevant part of a potentially much longer story (see Hutto 2004, 563–4; 2008a, 11–2). In line with the polite etiquette of conversational implicature reason explanations are therefore generally extremely compressed, truncated and elliptical.

The question of how to define ‘narrative’ has received a lot of attention (see Ryan 2007 for an overview). But it is not necessary to decide this issue in order to articulate and promote the NPH. The kind of narratives it requires do have a few, non-negotiable properties. They must be complex linguistic representations that describe a particular temporal sequence of (non-logical) related events (at least more than one, but usually a whole bunch). The event-sequences that narratives describe can be real or fictional. In this respect, the central construct of ‘narrative’ that I rely on is at least as well-defined as ‘theory’ or ‘simulation’.

Nelson supplies details of how children engage with stories and the developmental importance of this activity in fostering their cognitive skills during early childhood (2007, 172–8). Indeed, in light of the critical role that narrative practices play in this regard she notes ‘it remains surprising that developmental psychologists have not thus far systematically addressed the problem of how young minds come to understand stories’ (2007, 174).

It is interesting to note in this context that, ‘The two branches of cognitive science that concern themselves with practice – implementation of AI expert systems and empirical studies of experts – agree that the competence that underpins expert practice is not theory-like’ (Ohlsson 2007, 20).

Not all simulationists accept the claim that simulation is best understood in terms of projection (see Gordon 2005).

The idea that the psychological mechanisms responsible for this competence literally contain the principles in question I call NeoCartesianism, following Fodor (1983).

Contrast this diluted notion of what a ‘theory’ is with the robust one promoted by (Gopnik and Meltzoff 1997, 32–41). The formal variant is diluted precisely because it makes no claims about what is causally responsible for the existence of the patterns associated with the exercise of our FP-competence.

The implications of this are of major philosophical importance, for if it should turn out that the NPH is the best explanation of our FP-competence then the source of our intuitions about everyday psychological practice are not grounded in a shared ‘folk theory’ (Jackson 1998, 31–7). If so, attempts by philosophers to justify certain prevalent assumptions about the nature of the mind by appeal to commonsense will lack the kind of support they are often claimed to have.

The NPH abandons another of TT's central claims. For, originally TT was ‘not just the claim that there are principles which describe the behaviour of mental states. What is meant in addition to this is that mental states are what philosophers call “theoretical entities”. That is: it is not just that mental states are describable by a theory, but also that the (true, complete) theory of mental states tells us everything there is to know about them' (Crane 1995, 63). Accordingly, an implicit functional definition, in terms of roles mental states play in the wider mentalistic network was thought to constitute a ‘theoretical definition’ of, say, the concepts of belief and desire. And that sort of definition, as Crane notes, was meant to be exhaustive. But the claim that our understanding of the meaning of the concepts of the attitudes is exhausted by an understanding of their roles with respect to one another seems manifestly untrue. Children exhibit a practical understanding of these mental states and others prior to understanding their roles in reason explanations (see Hutto 2008a, chaps. 2, 6). Accepting this is wholly in line with the NPH but not certain prominent versions of TT.

Tying the invocation of simulative abilities to deep and shallow forms of narrative understanding, Kieran suggests that early story-telling encounters are shallow in important respects (2003, 70). Thus he observes that ‘Our narrative understanding can also become increasingly refined by coming to apprehend expressive qualities in terms of newly acquired concepts or ever finer distinctions amongst concepts we possess’ (2003, 77).

Currie provides an interesting discussion of the kind of framing activity, common to other communicative acts, that surrounds our understanding and appreciation of narratives in a recent paper (2007).

It is much less plausible that we must invoke simulative abilities in order to understand a rich literary description in a text that lies before us. For example, we seem perfectly capable of digesting the content of Dickens' depiction of Gradgrind without necessarily running any imaginative simulations at all, though doubtless we often do as a matter of fact (see Goldman 2006, 289).

In accepting this one can remain neutral on many other controversial theories about the nature of thoughts and thought-processes. For example, to accept this neither requires commitment to the existence of mental representations of any specific type (or even per se) nor to the claim that the ability is hardwired or sponsored by inherited mechanisms for intellectual reasoning. Thus the idea that we are capable of co-cognition is entirely compatible with the view that natural language is our ‘language of thought’ and that a capacity for sophisticated logical reasoning is late-developing, soft-wired and virtual.

It should be evident, pure co-cognition is not sufficient for making sense of the action of others. Failure to recognize this leads one to conflate thoughts with reasons (Stueber 2006, 156). Noting this, Heal is very clear about the limits of her simulative proposal. She observes that her co-cognitive account is about predicting the outcomes of certain thought processes and is therefore not in the business of accounting for how we go about ‘interpreting and explaining behaviour’ (Heal 1998a, 86). If co-cognition used principles it would need only one: ‘When an agent A acquires the belief that p and a rational thinker ought to infer q from the conjunction of p with other beliefs that A has, A comes to believe that q’ (Botterill 1996, 116).

Some suggestive research has been done on this topic (Lillard 1997, 1998; Vinden 1996, 1999, 2002, 2004). More extensive and targeted work is required in order to make the relevant cross-cultural comparisons in ways that would matter for the NPH. Some proposals to do just this are being currently developed. For example, Sinha, as principal investigator, has recently developed two major research grants which seek to examine the narrative practices in an Amazonian culture. These explicitly incorporate attempts to explore and test the NPH.

Some concerns of this sort were raised by Alvin Goldman in his presentation to ‘The Narrative Alternatives to Theories of Mind’ conference, held in July 2007 at the University of Hertfordshire.

Goldman has approached this task with gusto because he conceives of ‘ST as a scientific hypothesis, intended to compete on empirical terms with other cognitive scientific theories of mentalizing’ (2006, 20).

This should make us wary of the alarming tendency amongst contenders in the TT–ST debate to talk in almost pre-positivist terms of evidence speaking ‘for’ or ‘in favour’ of one theory or another. This is allied with a related tendency of theorists to actively seek evidence that supports their favoured hypothesis rather than looking for ways to falsify it. As such, a canny evaluator must be on the lookout to see if they have fallen foul of confirmation biases in reviewing their efforts. In discussing a typical case of this kind, Stueber rightly notes that ‘these results do not speak unequivocally for theory theory and against simulation theory … it appears that simulation theorists could claim that these findings are compatible with their position. It certainly does not speak against the claim that understanding others’ thoughts as reasons we have to conceive of them in a manner such that we could understand them as our reasons' (Stueber 2006, 168).

Notice too that the tasks in question are all cases of third-personal predictions about mental states. It seems even if we restrict ourselves to cases of third-person ‘mindreading’ the existing scientific data runs out of steam; as Goldman confesses: ‘I know of no theoretical analysis or experimental evidence that bears directly on simulation's role in retrodictive mental attribution’ (2006, 184).

This may not be transparently obvious for when it comes to investigating FP-competency, the empirical situation is made even more murky by certain default interpretations of ‘the data’ – interpretations that are not philosophically innocent (see Hutto forthcoming).

For a discussion of issues surrounding the evolutionary origins of our FP-competence, see Hutto (2008a, chaps. 11, 12; 2008b).

Additional information

Notes on contributors

Daniel D. Hutto

Email: [email protected]