An Investigation of the Perruchet Effect

Abstract

The single versus dual processing systems debate is one that has taken centre stage in the human learning literature. The existence of a propositional reasoning system is not disputed in this thesis, but whether a secondary processing system is required is. This is specifically tackled by investigating the mechanisms which underlie the Perruchet effect, an effect which is used widely to support a dual processing systems stance. During the Perruchet paradigm a single conditioned stimulus (CS) is partially reinforced by an unconditioned stimulus (US). Conditioned responding is found to dissociate from conscious expectation of the US across runs of reinforced (CS-US) and non-reinforced (CS-noUS) trials. US expectancy ratings typically fluctuate in accordance with the gambler’s fallacy. Conversely associative mechanisms are postulated to govern the variable strength of the conditioned response (CR). The associative nature of the CR is the subject of this thesis as it is queried whether a non-associative mechanism might explain this result. Three different methodological strands of the Perruchet effect are studied in this thesis: autonomic conditioning (Chapters 2 and 3), eyeblink conditioning (Chapter 3) and reaction time (RT) studies (Chapters 4 and 5). Additionally transcranial magnetic stimulation (Chapter 5) and computational modelling (Chapter 6) are used as tools to investigate the CR.

It is concluded in this thesis that the associative explanation of the CR in the Perruchet effect cannot be dismissed, although the strength of such an effect has perhaps been overstated in previous research. Evidence from autonomic conditioning provides the strongest evidence for an influence of CS-US association in the Perruchet effect as removal of the CS abolishes the CR in this thesis (Chapters 2 and 3). However, evidence from the eyeblink (Chapter 3) and RT (Chapters 4 and 5) variants of the effect suggest that there is undoubtedly a non-associative contribution to these effects. Although the exact mechanistic nature of this non-associative mechanism is unknown, priming is given as a possible explanation, and it is confirmed that such effects cannot be explained propositionally (Chapter 5). Overall a single processing system explanation of learning is not sufficient to explain the Perruchet effect.