Neuropsychologia 46 (2008) 1429–1441 Self-serving confabulation in prose recall Aikaterini Fotopoulou a,* , Martin A. Conway b , Mark Solms c , Stephen Tyrer d , Michael Kopelman a a King’s College London, Institute of Psychiatry, UK b Institute of Psychological Sciences, University of Leeds, UK c Departments of Neurology & Psychology, University of Cape Town, South Africa d Department of Psychiatry, University of Newcastle-upon-Tyne, Newcastle Upon Tyne, UK Received 22 August 2007; received in revised form 4 November 2007; accepted 20 December 2007 Available online 19 January 2008 Abstract Recent studies suggest that the content of confabulation is mainly positive and self-enhancing. In this group study, we aimed to investigate whether this positive bias is specific to self-referent information. Confabulating amnesic patients, amnesic non-confabulating patients and healthy controls were asked to reproduce a series of short stories. We manipulated the emotional valence of the material by including positive, negative and neutral story plots. We also manipulated the self-reference of the material by including self-referent versus other-referent encoding instructions. Confabulating patients were as impaired as a group of amnesic patients in the amount of information they recalled, both groups being worse than healthy controls. Importantly, confabulating patients showed a selective bias in the negative self-referent condition, in that they recalled such information in a manner which portrayed a more positive image of themselves. This positive bias was not present in stories that were not encoded in a self-referent manner and it was not significantly correlated to patients’ self-reported mood. We propose that both confabulation and its motivated content result from a deficit in the control and regulation of memory retrieval, which allows motivational factors to acquire a greater role than usual in determining which memories are selected for retrieval. To this extent, the self-enhancing content of confabulation could be explained as a neurogenic exaggeration of normal self-serving memory distortion. © 2008 Elsevier Ltd. All rights reserved. Keywords: Confabulation; Emotion; Memory; Valence; Limbic; Prefrontal cortex 1. Introduction Human memory is capable of storing previous experience and information accurately and over long periods, but it is far from a perfect system. On the contrary, memory can be subject to errors of both omission and commission and thus recollection may include instances of forgetting, or distorting past experience. The investigation of pathological forms of forgetting, such as amne- sia, has proven useful in understanding the nature of human learning (e.g. Milner, 1966). Similarly, the study of patholog- ically exaggerated forms of memory distortion has provided insight into the mechanisms of normal reconstructive remem- * Corresponding author at: Academic Unit of Psychiatry, St. Thomas’s Hos- pital, South Wing, Block 8, London SE1 7EH, UK. Tel.: +44 207 1880204; fax: +44 207 6330061. E-mail address: a.fotopoulou@iop.kcl.ac.uk (A. Fotopoulou). bering (Johnson, 1991; Kopelman, 1999; Schacter, Norman, & Koutstaal, 1998). An extreme form of pathological memory dis- tortion is confabulation, i.e. the production of fabricated, or distorted memories about one’s self or the world, without the conscious intention to deceive. Advances in the study of con- fabulation have informed and enriched a number of theoretical accounts of memory (Burgess & Shallice, 1996; Conway, 2005; Gilboa et al., 2006; Johnson, 1991). Over the last two decades there is greater knowledge of the probable neuroanatomical basis of confabulation and its con- comitant cognitive deficits. Most recent studies converge on the finding that the production of confabulation is linked to lesions in the ventral and orbital regions of the frontal lobes (Gilboa et al., 2006; Johnson, Hayes, D’Esposito, & Raye, 2000; Schnider, 2003). These lesions seem to lead to cognitive deficits in higher- order memory processes, including the inability to (a) retrieve memories in an organised and goal-oriented way (e.g. Burgess & Shallice, 1996; Moscovitch, 1989); (b) distinguish between 0028-3932/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.neuropsychologia.2007.12.030