Conflict components of the Stroop effect and their a??controla??

摘要

The more frequent the incongruent trials are in an experiment, the smaller the Stroop effect (Logan and Zbrodoff, 1979; Logan, 1980; Tzelgov et al., 1992). This is known as the list-wide proportion congruent effect because the proportions are manipulated at the list level. The list-wide proportion congruent effect is considered to be a marker of control since it demonstrates the modulation of the magnitude of the Stroop effect, and as such, is interpreted in terms of conflict adaptation. According to the conflict-monitoring framework (Botvinick et al., 2001; De Pisapia and Braver, 2006; Braver, 2012), an increased proportion of incongruent trials results in higher conflict at the response level, which triggers the control system. The control process includes two stages: conflict detection and control exertion. Referring to our previous discussion, in order to claim the automatic reading process can be controlled, the TC should be the target of both stages of the control process. However, according to our analysis, this is not the case. In fact, the TC is only a target of the control exertion stage. According to the models within the conflict-monitoring framework, conflict reduction is achieved through adjusting the weights of the two tasks, thereby minimizing the TC. However, the conflict detection stage is centered on response competition, which requires the TC to be amplified by the IC. When there is no IC, that is, no competing color-concept activation by reading, no competing color-response can be activated. Focusing on response competition (and thereby on IC) by Botvinick et al. and later models (De Pisapia and Braver, 2006; Blais et al., 2007; see also Verguts and Notebaert, 2008, for a model integrating cognitive control and reinforcement learning) leads to the conclusion that the detection of conflict, and therefore triggering of the entire control process, requires the IC being present (see Kalanthroff et al., 2013b, for evidence inconsistent with this assumption). There is no “path” in these architectures allowing TC to be monitored without the presence of the IC (Figure ​(Figure1).1). That is, the theory behind these architectures in their current state does not allow an unequivocal claim that reading, as an automatic process, can be controlled.