Theory of mind and language evolution: an exploration of rapid and involuntary perspective-taking

Abstract

Current research on language evolution has provided considerable insight into the emergence of linguistic structure, but much of this research does not account for how language users access speaker meanings to build mappings between meanings and signals. Signal-meaning mappings may arise in natural communication systems by three main routes: signal first, meaning first, or signal and meaning simultaneously. Humans seem uniquely capable of creating new simultaneous signal-meaning pairs as a result of our ability to make our communicative intentions apparent. This "ostensive-inferential'' communication can explain how meanings are attached to novel signals in infancy, emerging languages, and everyday communication by appealing to mindreading, or theory of mind – the ability to infer the intentions, knowledge, and other mental states of interlocutors.

The ostensive-inferential model of communication has been criticised on the grounds of implausibility; specifically, that neither infants nor adults are capable of the kind of rapid and complex mindreading required by this model. However, a range of experimental paradigms appear to show evidence of rapid and unconscious mindreading in both adults and infants. One paradigm in particular, the Dot Perspective Task (DPT), has been argued to show evidence of "spontaneous" or "automatic" mindreading in adults, although this interpretation is the subject of considerable dispute, with some evidence suggesting that the results are best explained by cognitive processes that do not involve mindreading.

In this thesis, I present a range of experiments using the Dot Perspective Task to investigate whether the classic result in this paradigm is best explained by mindreading or by alternative explanations. Using an adapted and extended set of DPT stimuli, I first investigate whether the "mindreading" effect is found only for human-like avatars, or whether it extends to non-human arrows. An exploratory analysis of the data from this task suggests that the results may be best explained by a spatial confound in the stimuli.

A subtle but critical difference in the implementation of DPT variants in the literature may explain not just this result, but other irreconcilable differences in the DPT literature. In a sequence of five experiments, I investigate this variation in experimental method. The results from these tasks suggest that the DPT does not demonstrate automatic mindreading – that is, mindreading that is reflexive and purely stimulus driven; but rather spontaneous mindreading, or mindreading that is rapid, unconscious and involuntary, but directed by attentional systems. This finding prompts us to clearly distinguish spontaneity from automaticity.

I present two further experiments investigating the mechanism underlying the classic DPT result. An adaptation of the DPT produced null results, as did a simplified version of the same task. This null result may be explained by the increased demand of the task, but given inconsistent findings across the DPT literature, it may also be the case that the classic DPT finding is less robust than it appears to be. I therefore review the current literature on failed replications in psychology and other disciplines, identifying how the problems described in this literature are relevant to the Dot Perspective Task. I argue that the use of the DPT to investigate rapid and involuntary mindreading remains promising, but that drawing firm conclusions from this paradigm will require replications of the current results with sufficient statistical power, as well as a thorough investigation of the wide range of methodological implementations that may affect the results of the task.

Finally, I explore an alternative route for future investigation of the plausibility of the ostensive-inferential model of communication. I discuss current attempts to reconcile the ostensive account with empirical research on mindreading, arguing that "minimalist" accounts of ostension fail to capture some of the central features of ostensive communication. I then suggest a reformulation of ostension that draws on the concepts of joint attention, common ground and mutual manifestness to offer an account of ostension that is both developmentally and cognitively more plausible than an account that relies on mindreading. I suggest that joint attention would be a fruitful avenue for future research on ostension.