Behavioural Brain Research 236 (2013) 332–343 Contents lists available at SciVerse ScienceDirect Behavioural Brain Research j ourna l ho me pa ge: www.elsevier.com/locate/bbr Research report Event-related potential correlates of emergent inference in human arbitrary relational learning Ting Wang, Simon Dymond ∗ Department of Psychology, Swansea University, Singleton Park, Swansea SA2 8PP, United Kingdom h i g h l i g h t s ◮ Examined functional–anatomical correlates of emergent relational inference. ◮ Relations trained between either words and pseudowords or arbitrary symbols. ◮ EEG was recorded during presentations of related and unrelated stimulus pairs. ◮ Faster, more accurate responses on symmetry and equivalence trials. ◮ ERPs were significant at mainly frontal–parietal and occipital sites. a r t i c l e i n f o Article history: Received 2 July 2012 Received in revised form 17 August 2012 Accepted 21 August 2012 Available online xxx Keywords: Categories Concepts Relational inference Stimulus equivalence Relatedness Directly trained Event-related potentials a b s t r a c t Two experiments investigated the functional–anatomical correlates of cognition supporting untrained, emergent relational inference in a stimulus equivalence task. In Experiment 1, after learning a series of conditional relations involving words and pseudowords, participants performed a relatedness task during which EEG was recorded. Behavioural performance was faster and more accurate on untrained, indirectly related symmetry (i.e., learn AB and infer BA) and equivalence trials (i.e., learn AB and AC and infer CB) than on unrelated trials, regardless of whether or not a formal test for stimulus equivalence relations had been conducted. Consistent with previous results, event related potentials (ERPs) evoked by trained and emergent trials at parietal and occipital sites differed only for those participants who had not received a prior equivalence test. Experiment 2 further replicated and extended these behavioural and ERP findings using arbitrary symbols as stimuli and demonstrated time and frequency differences for trained and untrained relatedness trials. Overall, the findings demonstrate convincingly the ERP correlates of intra- experimentally established stimulus equivalence relations consisting entirely of arbitrary symbols and offer support for a contemporary cognitive-behavioural model of symbolic categorisation and relational inference. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Traditionally, psychologists and philosophers alike have been interested in the seemingly unique human propensity to classify, categorise and organise linguistic stimuli. Categorisation and con- cept formation abilities have long been considered to be defining features of symbolic behaviour that often cannot readily be traced to a history of direct learning. Early behavioural approaches, for instance, emphasised the role of principles of reinforcement, dis- crimination and generalisation in concept learning based on direct learning [1]. Recently, behavioural psychology has developed a fruitful and rigorous approach to the study of categorisation and ∗ Corresponding author. Tel.: +44 1792 295602; fax: +44 1792 295679. E-mail addresses: t.wang@swansea.ac.uk (T. Wang), s.o.dymond@swansea.ac.uk (S. Dymond). symbolic behaviour, called derived relational responding, which is based on Sidman’s [2] stimulus equivalence paradigm. Historically, the phenomenon of stimulus equivalence dates back to ancient Greece [3] and was studied by experimental psychologists, such as stimulus–response (S–R) theorists [4] for decades until the demise of S–R psychology [5]. However, it was not until the early 1970s that Sidman rediscovered the topic and set about devising a coher- ent set of experimental procedures and terminology with which to study it [6]. Research on stimulus equivalence and other forms of derived relational responding has generated considerable interest because it may provide a novel approach to the investigation of unlearned or emergent categorisation skills involving physically distinct, arbitrarily related stimuli. The basic finding shows that when verbally-able humans learn a series of interconnected conditional discriminations, the stimuli often become related to one another in ways not explicitly trained. For instance, if choosing Stimulus B in 0166-4328/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbr.2012.08.033