Appetite (2002) 39, 119±125 doi:10.1006/appe.2002.0493, available online at http://www.idealibrary.com on 1 Drinking while thirsty can lead to conditioned increases in consumption P. J. Durlach a , N. A. Elliman b , and P. J. Rogers c a Army Research Institute, Orlando, Florida, USA; b Department of Psychology, Aston University, Birmingham, UK and c Department of Experimental Psychology, University of Bristol, Bristol, UK (Received 2 October 2001, revision 10 April 2002, accepted in revised form 16 April 2002) A within-subject design was used to test whether repeatedly drinking a novel-flavoured and coloured drink while thirsty would influence subsequent liking for or consumption of that drink, compared to a different flavoured and coloured drink repeatedly consumed while less thirsty. Each participant was given 300 ml of one flavoured drink (H) after consuming a high salt meal (527 g of salt), and 300 ml of another flavoured drink (L) after consuming a low salt meal (127 g of salt). Participants had 4 sessions with each meal-type/drink combination, in an intermixed order. Pre- and post-training assessments of the drinks were conducted to determine the impact of the training regime on pleasantness and perceived thirst-quenching effect of the drinks. The final session included a choice test, and ad libitum access to the chosen drink, after either a high or low salt meal. In this final choice session, people drank almost twice as much H as L; however, there were no differential effects of past training on rated liking or choice. The increased consumption of H might reflect greater liking for H which was not detected by the rating scales; or it might reflect the learning of greater ``conditioned thirst'' in response to the flavour of H. # 2002 Elsevier Science Ltd. All rights reserved. Liking of a food or beverage is one of the primary factors influencing food choice (e.g. Rozin & Schulkin, 1990; Steptoe et al., 1995). The determinants of liking are varied, including individual differences in sensory perception, opportunity for exposure determined by prevailing cultural/social norms, and individual ex- periences with the food (e.g. Mela, 2001). With respect to the latter, it has been found that the post-ingestive effects of the food can have consequences for later preference and consumption (e.g. Booth et al., 1982; Johnson et al., 1991; Sclafani, 2001). It is hypothesised that positive post-ingestive effects promote liking for the flavour/texture of the food or drink that precedes them, and that negative post-ingestive effects promote aversion, through some kind of associative process. While the results of animal studies clearly demonstrate such effects (Behmer et al., 1999; Garcia et al., 1976; Gibson & Booth, 2000; Kyriazakis et al., 1999; Sclafani, 2001), analogous evidence in humans has been relatively less clear. While it has been shown that children come to prefer flavours associated with high energy content (Birch et al., 1990; Johnson et al., 1991; Kern et al., 1993), Zandstra (Note 1) failed to find analogous effects in adults. Other studies have shown that caffeine consumers come to prefer flavoured drinks that contain caffeine. However, this occurs only if consumers are caffeine-deprived during learning, and appears to be expressed more strongly if consumers are also caffeine-deprived during testing (e.g. Rogers et al., 1995; Yeomans et al., 2000). Interestingly, Yeomans et al. (2000) and Yeomans et al. (in press) have also shown that caffeine con- sumers learn a mild aversion to the flavours of noncaffeinated drinks that they experience repeatedly while caffeine deprived; this aversion is analogous to the caffeine-conditioned preference, in that it too, is only observed while the participant is caffeine- deprived. Thus physiological state can modulate the expression of learned likes and dislikes. This research was conducted at the Institute of Food Research, Reading, and was supported by a grant from Unilever PLC. Thanks are given to David Mela, Hilde Hendrickx, and Martin Yeomans for their comments on an earlier draft of this manuscript, and to Henk Smit for conducting the unpublished study referred to in the text. Address all correspondence to: P. J. Durlach, Army Research Institute, Simulator Systems Research Unit, 12350, Research Parkway, Orlando, FL, 32826, USA; E-mail: Pjdurlach@hotmail.com 0195±6663/02/$ ± see front matter # 2002 Elsevier Science Ltd. All rights reserved.