Increasing set size breaks down sequential congruency: Evidence for an associative locus of cognitive control ☆ Chris Blais a, 1 , Tom Verguts b, ⁎ , 1 a University of California, Davis, United States b Ghent University, Belgium abstract article info Article history: Received 27 October 2011 Received in revised form 6 June 2012 Accepted 7 July 2012 Available online xxxx PsycINFO codes: 2300 2340 4160 Keywords: Cognitive control Associative learning Computational modeling In recent years, a number of studies have revealed that cognitive control is strongly context-dependent (e.g., Crump et al., 2006). Inspired by this, computational models have been formulated based on the idea that cognitive control processes are based on associative learning (Blais et al., 2007; Verguts & Notebaert, 2008). Here, we test a natural consequence of this idea, namely, that sequential congruency effects (Gratton et al., 1992) should gradually decrease with an increasing number of task-relevant features (e.g., stimuli). The effect is empirically observed and simulated in a computational model. Implications of our findings are discussed. © 2012 Elsevier B.V. All rights reserved. 1. Introduction In order to be useful, cognitive control must act fast. Consider the situation in which roadwork forces you to avoid construction workers on your habitual route to work. If your control system worked only at slow time scales, you would not be able to avoid them, leading to great danger for all persons involved. Consistent with this example, recent studies have emphasized that control is flexible and fast. A key phenomenon motivating this dynamic view of control is the sequential congruency effect (SCE; Gratton, Coles, & Donchin, 1992). Consider for example the Stroop task, where the task is to name the ink color of words. Stimuli can be either congruent (e.g., RED written in red) or incongruent (e.g., RED written in blue). The congruency effect is the difference in performance (e.g., reaction time) to incon- gruent versus congruent stimuli (see MacLeod, 1991 for a review). The SCE refers to the observation that the congruency effect is re- duced following an incongruent trial compared to a congruent trial. This effect demonstrates that control does not operate (only) at slow time scales, but on a trial-by-trial basis. In the present paper, we investigate the nature of cognitive control as pertains to the SCE in an experiment and a computational modeling study. A central construct to explain how a cognitive system can deploy fast control is response conflict. Botvinick, Braver, Barch, Carter, and Cohen (2001) introduced this concept to their conflict monitoring model to account for the SCE (in addition to explaining traditionally “slower” cognitive control manifestations such as listwise proportion congruency effects; Tzelgov, Henik, & Berger, 1992). Response conflict occurs when two or more competing responses are simultaneously active (see Appendix and Botvinick et al. for formal definition). They assumed that when response conflict is detected on a given trial, the current task (e.g., color naming) is given extra attention. Formally, this was implemented by increasing activation of a task demand unit (Cohen, Dunbar, & McClelland, 1990) that codes for the current task (e.g., “name ink color”) and that biases processing of the input dimen- sion (e.g., color). Because there is more response conflict on an incon- gruent trial than on a congruent trial, there is also more attention to the relevant input dimension (e.g., color) after an incongruent trial, leading to the SCE. Despite its success, the conflict monitoring model has difficulty dealing with a number of recent findings (Blais, Robidoux, Risko, & Besner, 2007). Consider again the Stroop task. Suppose that colors red and green are presented mostly congruently and colors blue and yellow mostly incongruently. Jacoby, Lindsay, and Hessels (2003) showed that the congruency effect is then larger for red and green. Acta Psychologica 141 (2012) 133–139 ☆ CB was supported by a postdoctoral fellowship from the Natural Sciences and Engineering Research Council of Canada (NSERC). TV was supported by Ghent University BOF/GOA, BOF08/GOA/011. ⁎ Corresponding author at: Department of Psychology, H. Dunantlaan 2, 9000 Ghent, Belgium. Tel.: +32 9 264 64 08; fax: +32 9 264 64 96. E-mail addresses: chris.blais@gmail.com (C. Blais), Tom.Verguts@Ugent.be (T. Verguts). 1 Both authors contributed equally to this paper. 0001-6918/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.actpsy.2012.07.009 Contents lists available at SciVerse ScienceDirect Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy