Action-based and vision-based selection of input: Two sources of control

In the first part of this paper we review evidence suggesting that there exists a mechanism that selects input on the basis of its similarity to the required action. This response-based input selection differs from the more established space- and object-based input selection in that it is not constrained by the structure of the input. Our evidence suggests that the two-choice Stroop effect is caused by this response-based selection mechanism. By contrast, it is known that the flanker effect is determined by the space- and object- based selection mechanisms. We explore whether the conflict resolution of the Stroop and flanker tasks is different as well by embedding these two tasks in a PRP (Psychological Refractory Period) paradigm. We show that the Stroop and the PRP effects are additive whereas the flanker and the PRP effects are underadditive, suggesting that the processes in charge of the conflict resolution in the Stroop and the flanker tasks are indeed different. We discuss possible reasons for this difference, and discuss possible ways in which the response-based mechanism can be implemented in information processing models. Control of action, the major theme of this special issue, can be exerted at several different levels of behavior. One type of control concerns input selection. Selecting a portion of input information for further processing, generally assumed to be done by attentional mechanisms, is a fundamental and integral part of task performance (e.g., Eriksen & Eriksen, 1974; Logan, 1996; Posner, 1980). It is commonly assumed that a major role of this selection is to filter out input stimuli that are irrelevant for the task at hand, although the precise locus at which irrelevant stimuli are blocked is still a subject of intense debate (e.g., Pashler, 1998). Two attentional mechanisms have been proposed for visual input selection. Space-based attention operates by specifying a region in space and selecting input located within that region. Object-based attention operates by selection of an object, and filtering out other objects. There is considerable evidence for the operation of both space-based (e.g., Eriksen & Yeh, 1985; Posner, 1980), and object-based (e.g., Duncan, 1984; Kramer, Weber & Watson, 1997) mechanisms. Indeed, there is evidence that both mechanisms may be operating simultaneously (Behrmann & Tipper, 1999). In the first part of this paper we focus on input selection. We review evidence from our lab suggesting that, in addition to the space- and object-based mechanisms (termed here interchangeably visual attention or vision-based mechanisms), there exists a third, response-based mechanism for input selection. A number of previous findings in the literature may be interpreted as stemming from the operation of this response-based mechanism of input selection. We point out that although the goal of this response-based mechanism is similar to those of space- and object-based, its mode of operation may be fundamentally different. Input selection is not always successful in blocking irrelevant input (i.e., distractors). Indeed, the very existence of interference paradigms such as the Stroop (e.g., MacLeod, 1991; Stroop, 1935) and the flanker (e.g., Eriksen & Eriksen, 1974) tasks is caused by inability to filter out the irrelevant written words and irrelevant flankers. In the second part of the paper we focus on the Stroop and the flanker interference paradigms. More specifically, we examine conflict resolution in the two tasks, required when the target and distractors are associated with different responses. Our studies (reviewed in the first part) suggest that the Stroop effect is caused by a failure of the response-based mechanism to filter out the words. Review of the literature suggests, in contrast, that the flanker effect is caused by a failure of the vision-based mechanism to block the flankers from affecting performance. We present novel experiments, using the PRP paradigm, demonstrating that the conflict resolution in these two paradigms is fundamentally different. In the discussion we focus on the relation between the selection mechanisms and conflict resolutions when selection fails.