PSYCHOLOGICAL SCIENCE Research Report 370 Copyright © 2002 American Psychological Society VOL. 13, NO. 4, JULY 2002 VISUAL SEARCH IS MODULATED BY ACTION INTENTIONS Harold Bekkering 1,2 and Sebastiaan F.W. Neggers 1,3 1 Department of Cognition and Action, Max Planck Institute for Psychological Research, Munich, Germany; 2 Department of Experimental and Work Psychology, University of Groningen, Groningen, the Netherlands; and 3 Helmholtz Institute, Unit of Psychonomics, Utrecht University, Utrecht, the Netherlands Abstract—The influence of action intentions on visual selection pro- cesses was investigated in a visual search paradigm. A predefined tar- get object with a certain orientation and color was presented among distractors, and subjects had to either look and point at the target or look at and grasp the target. Target selection processes prior to the first saccadic eye movement were modulated by the different action in- tentions. Specifically, fewer saccades to objects with the wrong orien- tation were made in the grasping condition than in the pointing condition, whereas the number of saccades to an object with the wrong color was the same in the two conditions. Saccadic latencies were similar under the different task conditions, so the results cannot be explained by a speed- accuracy trade-off. The results suggest that a specific action intention, such as grasping, can enhance visual processing of action-relevant features, such as orientation. Together, the findings support the view that visual attention can be best understood as a selection-for-action mechanism. Typically, a person who searches for a specific object in the envi- ronment also has a specific intention of what he or she wants to do with that object. For instance, someone might search for a big red coffee cup with either the intention to grasp it and subsequently drink from it or the intention to point to it to communicate to a colleague that he or she should use that cup to have some coffee. It is important to note that the physical characteristics of the cup are the same (e.g., big and red) for the two tasks. Nevertheless, information about size and orientation of the cup is irrelevant to pointing to it, whereas such information is very relevant for the grasping action. For example, early movements aimed at grasping an object are influenced by the object’s orientation (Ma- massian, 1997), but pointing movements are not (Smeets & Brenner, 1999). Therefore, an efficient visual search system should detect action-relevant information in an enhanced way. Some first evidence for the existence of an action-related attentional search system has recently been reported. For instance, Humphreys and Riddoch (2001) found that a patient with symptoms of unilateral ne- glect was slow and sometimes unable to find targets when they were defined by their name or even by a salient visual property (such as their color), but he was relatively efficient at finding a target defined by the action it afforded. In addition, Meegan and Tipper (1999) found that in- terference from a distractor was uniquely related to a reaching task and did not interfere when the same setup was used in a verbal task. The goal of the present research was to investigate further action-related se- lective visual attention in visual search. Selective visual attention, the mechanism that enables an organism to select specific information that is registered by the senses for further processing and visually guided behavior, is currently thought to be con- trolled by two distinct mechanisms. On the one hand, attentional con- trol is said to be stimulus-driven when visual selective attention is activated in a bottom-up manner. In this case, the strength of bottom-up activation, which is based on the differences between neighboring items (Wolfe, 1994), guides attention toward distinctive items in the field (see also Duncan & Humphreys, 1989). On the other hand, attentional control is said to be user-driven when visual selective attention is activated in a top-down manner. In this case, the observer is thought to select a subset of the available visual features that corresponds to the recognition task at hand. User-driven visual se- lection can be best explained by the example of a conjunction search. In a conjunction search, the target might be specified as an object with two specific features; for example, the target might be a green cylinder, and distractors might be orange cylinders and green and orange rectan- gles. Objects with the same shape or the same color (not both) are more often erroneously looked at than objects that do not share a feature with the saccadic target (Findlay, 1997). It is therefore argued that top-down specification plays a role not only in the programming of saccadic eye movements but also in visual search in general (Cave & Wolfe, 1990). Remarkably, however, although attentional control is suggested to be user-driven, or goal-directed (Gibson, 1996), the interaction between the visual information and the actor has typically been minimized. Goal- directed visual processing has been investigated in cognitive tasks; for ex- ample, participants have been asked to report the presence of a certain item in a full display by selecting one of two possible button presses, an act that by itself does not require knowledge of an object property in order to be ex- ecuted (e.g., Treisman & Gormican, 1988). In other words, the top-down specification is set by the task instructions, and not by the action itself. The aim of this study was to investigate whether visual selective processes are modulated by the action intentions of the actor. Specifi- cally, the question was whether processing of task-relevant informa- tion is enhanced compared with processing of task-irrelevant information. In one condition, subjects were asked to saccade and point to a target object as quickly and accurately as possible; in the other condition, they were asked to saccade to and grasp the predefined target object. Orienta- tion, but not color, is known to influence the way an object is grasped. In contrast, pointing to a target is known to be insensitive to both ob- ject orientation and color (Smeets & Brenner, 1999). Therefore, fewer orientation errors were expected in the saccade-and-grasp condition than in the saccade-and-point condition, but the number of color errors was expected to be the same in the two conditions. METHOD Participants Participants had normal or corrected-to-normal vision and were tested to ensure they were not color-blind. The 8 participants, 4 males and 4 females, had a mean age of 25 years and were paid 15 Deutsch Marks. Address correspondence to Harold Bekkering, Experimental and Work Psy- chology, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen, The Netherlands; e-mail: h.bekkering@ppsw.rug.nl.