How Incidental Sequence Learning Creates Reportable Knowledge: The Role of Unexpected Events Dennis Ru ¨nger Berlin–Brandenburg Academy of Sciences and Humanities Peter A. Frensch Humboldt-Universita ¨t zu Berlin Research on incidental sequence learning typically is concerned with the characteristics of implicit or nonconscious learning. In this article, the authors aim to elucidate the cognitive mechanisms that contribute to the generation of explicit, reportable sequence knowledge. According to the unexpected- event hypothesis (P. A. Frensch, H. Haider, D. Ru ¨nger, U. Neugebauer, S. Voigt, & J. Werg, 2003), individuals acquire reportable knowledge when they search for the cause of an experienced deviation from the expected task performance. The authors experimentally induced unexpected events by disrupt- ing the sequence learning process with a modified serial reaction time task and found that, unlike random transfer sequences, a systematic transfer sequence increased the availability of reportable sequence knowledge. The lack of a facilitative effect of random sequences is explained by the detrimental effect of random events on the presumed search process that generates reportable knowledge. This view is corroborated in a final experiment in which the facilitative effect of systematic transfer blocks is offset by a concurrent secondary task that was introduced to interfere with the search process during transfer. Keywords: sequence learning, serial reaction time task, reportable knowledge, explicit knowledge, unexpected events Individuals can learn about the sequential structure of environ- mental events incidentally, that is, without prior intention to ap- prehend a sequential regularity. For the most part, empirical re- search on sequence learning has been concerned with the characteristics of implicit learning. Sequence learning is said to be implicit when it occurs in the absence of conscious or explicit knowledge about the sequential regularity (cf. Erdelyi, 2004). Little theoretical value has been attached to the ubiquitous finding that incidental sequence learning also creates explicit knowl- edge—at least in some participants. Moreover, a review of the literature suggests that the generation of explicit sequence knowl- edge varies systematically across experimental conditions (Fren- sch et al., 2003). For example, Frensch, Lin, and Buchner (1998; Experiments 2a and 2b) manipulated the amount of training, the type of training (i.e., single or dual task training), and the type of sequence to be learned with the serial reaction time (SRT) task (Nissen & Bullemer, 1987). On each trial with the task, a partic- ipant responds to a target that appears in one of several screen positions by pressing a spatially compatible response key. Re- sponse locations on consecutive trials conform to a fixed sequence that is continuously recycled throughout the training phase. All three factors were found to influence whether participants were able to provide a verbal description of the sequential regularity after the training phase. This and related findings beget the ques- tion of which cognitive mechanisms might be responsible for the generation of explicit knowledge during incidental sequence learn- ing. However, researchers have only recently begun to approach this particular issue theoretically; pertinent empirical research is still largely absent from the literature. The purpose of this report is to test a central prediction of a theoretical framework advanced by Frensch et al. (2003) to explain the acquisition of explicit, reportable knowledge in incidental learning situations. The framework was dubbed the unexpected- event hypothesis in reference to the central theoretical notion that unexpected events can trigger the generation of reportable knowl- edge. Before we delineate the hypothesis in greater detail, we review some alternative theoretical accounts that have been of- fered in the literature. It is possible to identify two broad classes of theories on the generation of explicit sequence knowledge—single-system and multiple-system accounts. According to the single-system view, implicit and explicit knowledge are rooted in the same set of learning mechanisms (e.g., Cleeremans, 2006; Cleeremans & Jime ´nez, 2002; Kinder & Shanks, 2003; Perruchet & Vinter, 2002; Shanks, Wilkinson, & Channon, 2003). In its most stringent form, the single-system view rejects the notion of separable knowledge bases altogether, that is, the distinction between implicit and explicit knowledge. It is assumed that all markers of learning, be it faster responses to sequentially structured stimuli or verbal descriptions of the sequential regularity, provide different expres- sions of the same underlying memory representations. Learning increases the quality of representations, which, in turn, leads to improved performance in all available measures of learning (e.g., Perruchet & Amorim, 1992; Perruchet, Bigand, & Benoit-Gonin, 1997). Dennis Ru ¨ nger, Berlin–Brandenburg Academy of Sciences and Human- ities, Berlin, Germany; Peter A. Frensch, Department of Psychology, Humboldt-Universita ¨t zu Berlin, Berlin, Germany. This research was supported by Federal Ministry of Education and Research Grant 01GWS061. Correspondence concerning this article should be addressed to Dennis Ru ¨nger, Berlin–Brandenburg Academy of Sciences and Humanities, Inter- disciplinary Research Group “Functions of Consciousness,” Ja ¨gerstr. 22/ 23, D-10117 Berlin, Germany. E-mail: dennis.ruenger@gmail.com Journal of Experimental Psychology: Copyright 2008 by the American Psychological Association Learning, Memory, and Cognition 2008, Vol. 34, No. 5, 1011–1026 0278-7393/08/$12.00 DOI: 10.1037/a0012942 1011