Long-term persistence of sort strategy in free classification Fraser Milton * , Andy J. Wills University of Exeter, Washington Singer Laboratories, Perry Road, Exeter EX4 4QG, United Kingdom article info Article history: Received 1 July 2008 Received in revised form 1 December 2008 Accepted 1 December 2008 Available online 12 January 2009 PsycINFO classification: 2343 Keywords: Family resemblance Free classification Perseveration Unidimensional abstract Two free classification experiments that investigate the persistence of sort strategy are reported. Partic- ipants tend to persist with their initial categorization type (family resemblance or unidimensional) for the remaining sorts, overriding the effects of otherwise influential stimulus properties. Sort type was found to persist even after a one-week delay. Stimulus-driven models of free classification (e.g., the SUS- TAIN model, [Love, B. C., Medin, D. L., & Gureckis, T. M. (2004). SUSTAIN: A network model of category learning. Psychological Review, 111, 309–332]) cannot predict the sort type persistence effects we observe, but they are naturally accounted for by theories that posit strategic selection of a problem-solv- ing strategy (e.g., Hypothesis theory, [Levine, M. (1971). Hypothesis theory and nonlearning despite ideal S-R-reinforcement contingencies. Psychological Review, 78, 130–140]). Ó 2008 Elsevier B.V. All rights reserved. 1. Introduction Classification – dividing the world into groups of things – is one of the fundamental building blocks of cognition. However, in view of the immense number of objects we encounter, this process must necessarily be highly constrained; for example, just 10 items can be partitioned in more than 100,000 different ways. Thus, a greater knowledge of how we acquire the categories we have is an impor- tant pre-requisite for our understanding of human cognition. The majority of research in this area has concentrated on a task where participants get accurate, trial-specific feedback about their categorization decisions, and the thorough study of this task has yielded important information. However, it is increasingly being appreciated that the level of feedback present in this task seldom exists outside the laboratory. As a consequence, there has been growing interest in how people categorize stimuli in the absence of feedback, a task described as free classification (Regehr & Brooks, 1995), spontaneous categorization (Pothos & Chater, 2002), or cate- gory construction (Medin, Wattenmaker, & Hampson, 1987). One of the fundamental phenomena in free classification is that people tend to exhibit a strong preference to sort on the basis of a single stimulus dimension (e.g., Medin et al., 1987), for example, sorting entirely on the basis of color, whilst apparently discounting variability on other dimensions (e.g., size). This result is perhaps surprising given that many natural categories seem to have a ‘‘fam- ily resemblance” (overall similarity) structure (Rosch & Mervis, 1975), in other words, they possess a number of characteristic but not defining features. Recent research has demonstrated that people’s preference for unidimensional sorting can be substantially reduced, and family resemblance sorting increased, by a number of different factors. For example, the prevalence of family resemblance sorting is in- creased by sequential stimulus presentation (Regehr & Brooks, 1995), by spatial separation of the stimulus dimensions (Milton & Wills, 2004), and by stimulus dimensions which are easier to perceptually differentiate (Milton & Wills, 2008). In addition, fam- ily resemblance sorting can be elevated by increasing the time available to categorize the stimuli (Milton, Longmore, & Wills, 2008), and by relevant background knowledge which allows fea- tures to be integrated in a meaningful and coherent way (Spalding & Murphy, 1996; Wattenmaker, 1995). The current article exam- ines another factor that might be expected to influence the preva- lence of family resemblance sorting – the sort type the participant has previously produced. Our hypothesis is that the prevalence of family resemblance sorting will be affected by the sort strategy employed previously in similar situations. This hypothesis seems generally in line with a number of previously reported findings which demonstrate that people commonly persist with an initial strategy, even when a re- lated task can be completed more effectively with a different strat- egy. The prototypical example of this is perhaps the water jug Einstellung problems used to study the perseveration of mental set (e.g., Luchins, 1942; Schultz, Stone, & Christie, 1997). Cognitive rigidity has also been demonstrated in a sequential number reduc- tion task (Woltz, Gardner, & Bell, 2000), the sequential application 0001-6918/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.actpsy.2008.12.001 * Corresponding author. Tel.: +44 1392 264636. E-mail address: f.n.milton@exeter.ac.uk (F. Milton). Acta Psychologica 130 (2009) 161–167 Contents lists available at ScienceDirect Acta Psychologica journal homepage: www.elsevier.com/locate/actpsy