Behavioural Brain Research 222 (2011) 270–273 Contents lists available at ScienceDirect Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr Short communication Differential outcomes aid the formation of categorical relationships between stimuli Alexander Easton a,∗ , Simon Child a,b , Ginesa Lopez-Crespo c a Department of Psychology, Durham University, South Road, Durham, DH1 3LE, UK b School of Psychological Sciences, University of Manchester, Manchester, UK c Departmento de Psicología y Sociología, Universidad de Zaragoza, Zaragoza, Spain article info Article history: Received 6 July 2010 Received in revised form 14 March 2011 Accepted 15 March 2011 Available online 31 March 2011 Keywords: Learning Humans Outcomes abstract The differential outcome effect is when learning is enhanced through the application of different out- comes to different conditions of a task. Here we explore whether one difference in learning with differential outcomes is an enhanced categorisation of objects. We demonstrate that participants learn- ing conditional discriminations are better able to identify previously unpaired objects as belonging to the same category when differential outcomes were used in learning these stimuli. © 2011 Elsevier B.V. All rights reserved. The differential outcome effect is the enhanced rate of learning or accuracy in responding when each condition of a task is associ- ated with a unique outcome. The effect was first demonstrated in rats where conditional discriminations were taught with a sucrose reward for one condition and a food pellet reward for the other condition of the discrimination [1]. In comparison to a group of animals that received the same reward for correct responses (not associated with a particular condition of the task), the differential outcomes group showed an enhanced rate of learning for the dis- crimination. Similar DOEs have been seen in a range of tasks and a range of species (e.g., [2–4]). Work on DOEs has suggested that the effect is due to an expecta- tion of the unique outcome which is then directly associated with the object being discriminated (e.g., [5]). This was supported by more recent work in humans which showed that participants could transfer information learnt about a conditional discrimination to others associated with the same unique reward outcome, but could not make a similar transfer of information when all discrimina- tions were rewarded with the same outcome [6]. Similarly, whilst visuo-visual conditional discriminations in monkeys are impaired by disconnection of prefrontal cortex and inferior temporal cortex [7], the same sorts of discriminations are neither impaired by the same disconnection if each condition is associated with a unique outcome [8], nor does this pattern of lesions prevent the transfer of information between conditional discriminations [8]. These data ∗ Corresponding author. Tel.: +44 191 3343257; fax: +44 191 3343241. E-mail address: alexander.easton@durham.ac.uk (A. Easton). support the hypothesis that conditional discriminations require configural associations to be made between the conditional cue and the choice object, whilst the same discriminations in the pres- ence of unique outcomes could be solved by forming an associative expectation of the specific outcome, and this associative solution allows transfer of learning to other problems with the expectation of the same outcome [6,8]. This finding fits with other data sup- porting the hypothesis that the frontal cortex is required in setting task conditions when they cannot be identified through a simple association with the stimuli presented [8–10]. Although expectation of a unique outcome appears to explain the DOE, an alternative possibility is that all conditional discrim- inations are learnt by categorisation of the choice objects into categories identified by the cue objects. In contrast, conditional discriminations associated with unique outcomes might be cate- gorised by the outcome instead of the conditional cue. In this case the outcome might be more salient to the participant than the nature of the conditional cue and therefore the categories may be learnt more rapidly (displaying a DOE) when differential outcomes are used. It is also possible that this explanation might explain the dissociation in effect of crossed lesions of frontal and inferior tem- poral cortex in monkeys on conditional discriminations with and without differential outcomes. Differential outcomes procedures might improve performance of conditional discrimination learn- ing to ceiling levels. Now, surgical disconnection may not show an impairment because the impairment does not take performance below this ceiling. In an attempt to address this question, the present study specifically examined the effect of differential outcomes on cate- 0166-4328/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.bbr.2011.03.036