Cognition at the symbolic threshold: the role of abductive inference in hypothesising the meaning of novel signals

Abstract

Humans readily infer the meanings of novel symbols in communicative contexts of varying complexity, and several researchers in the field of language evolution have explicitly acknowledged that inference plays a key role in accounting for the evolution of symbolic communication. However, in this field at least, there has been very little investigation into the nature of inference in this regard. That is, evolutionary linguists have yet to address the following questions if we are to have a fuller picture of how humans came to communicate symbolically: 1. What kinds of inference are there? Specifically, i Diachronically, what forms of inference are comparatively simpler in evolutionary terms, and thus shared with a wider range of species? What forms of inference are more complex, and limited to humans or to us and our closest relatives? ii Synchronically, if humans are capable of several kinds of complex inference, how do we know which particular kind of inference is being applied in solving a given problem? 2. How do symbol-learning problems vary? Specifically, i What makes a particular symbol-learning problem more or less complex in terms of the kind of inference needed to solve it? ii How would the communicative context of our pre-linguistic ancestors have been different from that of a human child learning words from its linguistic parent? This dissertation takes a step towards answering these questions by investigating a little-known form of inference called `abduction' (or insightful hypothesis generation), which has thus far been wholly overshadowed in language evolution by a much better understood form called `induction' (or probabilistic hypothesis evaluation). I will argue that abduction and induction are both comparatively complex in the diachronic terms expressed above in 1.i, and while induction is useful in accounting for how modern children learn words from linguistic adults, abduction is more important in situations like those that would have faced our pre-lingistic ancestors as they first began to use symbols. That is, I will argue on both theoretical and empirical grounds that abductive inference was an evolutionary milestone as our ancestors crossed what Deacon (1997) calls the symbolic threshold.