Running Head: CONFLICT ADAPTATION WITHIN AND ACROSS TASKS In press (as of 2014), Psychological Research/Psychologische Forschung Generality and Specificity in Cognitive Control: Conflict Adaptation within and across Selective-Attention Tasks but not across Selective-Attention and Simon Tasks

To explain how cognitive control is modulated contextually, Botvinick, Braver, Barch, Carter, and Cohen (2001) proposed that detecting information-processing conflict attenuates the disruptive influence of information-processing conflicts encountered subsequently, by which time appropriate cognitive-control mechanisms already will have been engaged. This conflictadaptation hypothesis has motivated extensive programs of research while also attracting vigorous methodological critiques that highlight alternative accounts of trial n x trial n-1 sequential effects in cognitive-control tasks. Addressing those alternatives through precluding analyzing stimulus repetitions without creating any sort of confounds among any stimulus or trial characteristics, the present research observed significant conflict-adaptation effects within and across several selective-attention tasks. Moreover, across-task conflict-adaptation effects were largest when spanning tasks (i.e., a newly developed Stroop-trajectory task and a flanker task, which both require resolving conflict among stimulus elements) that presumably depend on the same mechanism of cognitive control (selective attention) than when spanning tasks that do not (i.e., the Stroop-trajectory task and a Simon task, the latter—but not former—of which requires resolving conflict between stimulus and response elements). These findings contribute to advancing beyond examining whether or not conflict adaptation exists to clarifying the conditions under which it is and is not observed. Conflict Adaptation Within and Across Tasks 3 Adapting the control of action and cognition to fluctuating environmental demands is a hallmark of effective human behavior. An influential account of the contextual modulation of cognitive control has been Botvinick, Braver, Barch, Carter, and Cohen’s (2001) conflictmonitoring theory, which proposes that signals of information-processing conflict, as when different processing streams implicate incompatible responses, engage cognitive control. Following that theory, detecting an incident of information-processing conflict should attenuate the disruptive influence of information-processing conflicts encountered subsequently, by which time appropriate cognitive-control mechanisms already will have been engaged. This conflictadaptation prediction has generated a substantial body of research on human behavior and brain function (for review, see Carter, & Krug, 2012), giving rise to novel conceptual approaches to understanding such disparate phenomena as racial and gender stereotyping (Kleiman, Hassin, & Trope, in press), affective and hedonic experiences (Braem, Verguts, Roggeman, & Notebaert, 2012; Van Steenbergen, Band, & Hommel, 2009, 2010), neural development (Waxer, & Morton, 2011), and relations among electrocortical and behavioral indices of cognitive control in largescale correlational studies (Clayson & Larson, 2011). Alongside these fascinating applications of conflict-monitoring theory, however, questions have persisted pertaining to whether the conflictadaptation effect itself is a basic principle of cognitive control or instead a methodological artifact, as discussed below. From the outset, research on conflict adaptation has taken steps to address alternative explanations. Gratton, Coles, and Donchin (1992) first demonstrated that increases in response time and decreases in response accuracy on incongruent relative to congruent trials are smaller following incongruent trials than following congruent trials of cognitive-control tasks, such as an Eriksen flanker task (Eriksen & Eriksen, 1974). That initial paper reported that the adaptation Conflict Adaptation Within and Across Tasks 4 effect it had identified was not moderated significantly by whether or not subjects viewed identical stimulus arrays across trials n and n-1, such as “HHSHH” on both trials, relative to “HHSHH” at trial n-1 and “SSHSS” at trial n (Gratton et al., 1992). In contrast, Mayr, Awh, and Laurey (2003) found that conflict-adaptation effects, also on the flanker task, were not observed when exact stimulus repetitions were removed from analysis, suggesting that repetition priming might be a more parsimonious account of their data than conflict monitoring theory. Addressing those conflicting findings, subsequent work expanded the number of stimuli used as flankers and as targets (Ullsperger, Bylsma, & Botvinick, 2005; see also Notebaert & Verguts, 2006). Significant conflict-adaptation effects were observed despite limiting analyses to trial pairs in which no target, distractor, or response elements repeated across successive trials (e.g., “77677,” followed by “33233”; Ullsperger et al, 2005). As noted by Schmidt and DeHouwer (2011), however, many conflict-adaptation studies with greater than two responses have created stimulus-contingency confounds. For example, in a flanker task with more than two responses, each central cue of a flanker trial would appear more often with identical flankers (because there is only one type of congruent array for that target, e.g., “77777”) than with any other stimuli (because there are many types of incongruent arrays for that target, e.g., “11711,” “22722,” “33733,” etc.). Those authors showed that eliminating stimulus-contingency confounds can eliminate the conflict-adaptation effect (Schmidt & DeHouwer, 2011). In our view, it is important to note that unconfounding all stimulus and response elements in a four-response design requires .75 incongruent trials and .25 congruent trials across, leading to unequal proportions of the different sequences that form the primary independent variable in conflictadaptation studies (e.g., .75 * .25 = .1875 incongruent following congruent trials; and .75 * .75 = .5625 incongruent following incongruent trials). Correlating a manipulation of trial sequences’ Conflict Adaptation Within and Across Tasks 5 levels of information-processing conflict with a manipulation of trial sequences’ frequencies makes it impossible to attribute any result solely to one manipulation or the other. Accordingly, novel tests of the conflict-adaptation hypothesis are needed that (a) eliminate exact stimulus repetitions without (b) introducing stimulus-contingency confounds, while (c) holding congruency rates to .50. Toward this end, our first experiment examined conflict adaptation in a four-response selective-attention task (a newly developed “Strooptrajectory” task) in which stimulus arrays were oriented vertically or horizontally, thereby allowing analysis of trial pairs associated with different responses and different stimuli. Our remaining two experiments addressed the above three criteria by interspersing different tasks across trials, as described next. In our view, and in accord with Egner (2008), a straightforward implication of Botvinick and colleagues’ (2001) conflict-monitoring theory is that conflict-adaptation effects should be observed across different tasks to the extent that performance at trials n and n-1 depends on the operation of a common component process of cognitive control. To the extent that a component process of cognitive control (e.g., selective attention) is needed to resolve information-processing conflict at trial n, engaging that process at trial n-1 should facilitate resolving informationprocessing conflict at trial n (whether or not same task is performed at trials n and n-1), thereby leading to attenuated neural signals of information-processing conflict and attenuated increases in response time and decreases in response accuracy as a result of encountering incongruent relative to congruent trials of cognitive-control tasks. Importantly, if such (conflict-adaptation) effects were to be observed despite differences in stimuli perceived and rules followed in different tasks across trials n and n-1, the results would appear impossible to attribute to stimulus repetitions or to any introduction of stimulus-contingency confounds. However, variability in Conflict Adaptation Within and Across Tasks 6 whether conflict-adaptation effects have been observed across different tasks (e.g., Freitas, Bahar, Yang, & Banai, 2007; Kan et al., in press; Kleiman et al., in press; Kunde & Wuhr, 2006) or have not been observed across different tasks (e.g., Akçay & Hazeltine, 2011; Funes, Lupiáñez, & Humphreys, 2010; Wendt, Kluwe, & Peters, 2006) has yielded uncertainty regarding the implications of those studies for understanding the contextual modulation of cognitive control. The present work aimed to help resolve this uncertainty by bringing under experimental control factors determining whether or not across-task conflict adaptation is observed. More specifically, given the general proposal (Egner, 2008; Funes et al., 2010) that across-task similarity may help determine whether or not across-task conflict-adaptation is observed, our final two experiments respectively manipulated two bases of across-task similarity, relating to the structure of stimuli encountered (i.e., whether trials of different tasks encountered at trials n and n-1 were comprised of single or multiple objects, Experiment 2) and to the nature of information-processing conflict (i.e., whether trials of different tasks encountered at trials n and n-1 entailed resolving conflict between stimulus elements or between stimulus and response elements, Experiment 3). A novel methodological property of these experiments is that each included three tasks that varied along the above-noted dimensions, thereby affording the first experimental tests of which we are aware of whether the magnitude of conflict-adaptation effects would be greater across some task combinations than across others. Clarifying the conditions under which the contextual modulation of cognitive control generalizes across tasks that utilize different stimuli would provide clear support for the conflict-adaptation hypothesis while also helping elucidate the processes by which such effects unfold. Conflict Adaptation Within and Across Tasks 7