In a wide range of circumstances, one is often faced with the task of visually searching for an object among an arrangement of other distracting objects. Visual search occurs, for example, when one has to find a car in a parking lot, or an abnormality in a radiographic image, or a camouflaged missile launcher. The most common indicator of visual search ability is the time to acknowledge manually the presence or absence of a target among distractor stimuli. Manual reaction times (MRTs) have inspired various theoretical ideas on attention and visual search. Despite their differences, the various ideas preserve the basic tenets of Treisman and Gelade’s (1980) feature integration theory. The theory postulates that early visual processing of a scene involves the parallel and pre-attentive extraction of fundamental features such as orientation, size, and colour from objects. Following this pre-attentive extraction stage, attention is used to recombine the features to form specific objects. During search, objects that are more salient among distractors are processed more quickly and “pop-out”. Pop-out is inferred from equally fast MRTs irrespective of the number of neighbouring distractors. For objects that are not easily discriminable, MRTs are more linearly dependent on the number of distractors and search is said to proceed in a serial manner.

Notwithstanding the benefits of MRTs in the acknowledgement of the presence or absence of a target among distractors, a limitation is that they reflect the end-product of a dynamic process that (a) can involve movement of the eyes to find the target  and that (b) can involve shifts in attention, with eye movements and without eye movements. A vast number of studies discount the role of eye movements by presenting the stimuli within a small visual angle and/or by presenting the stimuli very briefly. However, perhaps because of the proliferation of eye tracking systems, researchers have increasingly considered the role of eye movements. Studies that have addressed the eye movement aspect of visual search have typically measured eye movement indices alone, yet eye movements alone do not readily provide an indication of when an observer decides to acknowledge a target.

Recent findings from studies that consider eye movement indices together with MRTs indicate correlations between the variables. For instance, Zelinsky and Sheinberg (1997) measured eye movement indices (e.g., number of saccades and fixation duration) and MRTs under easy (i.e., pop-out) and difficult (i.e., non-pop-out) search conditions. MRTs correlated positively with the number of saccades initiated before a target was manually acknowledged. When evaluated across the two search conditions, 67% of the variance in MRTs could be attributed to the number of saccades initiated before a manual response was made. Not surprisingly, the relationship was stronger in the difficult search condition, which had greater variability in MRTs. For the same stimuli, the combination of the number of saccades and the latencies of initial saccade also correlated strongly with MRTs. In this case, a higher percentage (85%) of the variance in MRTs was attributable to the combination of the latencies of initial saccade and the number of saccades.

In real-world settings, a direct eye movement index of visual search is the latency to fixate the target when it appears among distractors. For example, when a police officer or a peace-keeping soldier is faced with a shoot/no-shoot situation an important factor in making a (manual) response is when the target person or target event was fixated. A similar argument can be made for other ill-posed real world problems (driving a car, piloting an aircraft, playing sports, etc). The present study was concerned with theoretical relationships between the latency to fixate and the latency to make a manual response to a target. A simple visual search paradigm is used.

A study by Binello et al. (1995) is of background interest for the present endeavour. They required two trained observers to respond manually to the presence/absence of a target in multi-stimulus displays. The stimuli were presented after the disappearance of a central fixation point. In addition to responding manually, the observers were required to fixate the target as quickly as possible and to maintain fixation until the target disappeared. With each data point representing a different search task, the results indicated strong positive correlations between median time to fixate (and to maintain this fixation) of the target (TMF) and median MRTs. The functions describing the relationships were similar for the two observers: (1) despite differences in their eye movement search strategies (i.e., large number of short fixations vs. few long fixations) and (2) despite the various contexts of visual search. For the two observers, 96% and 90% respectively of the variance in MRTs were attributable to median TMF. In addition, the linear regression lines had slopes of almost 1 with y-intercepts of 20 ms. Thus theoretically, MRTs occurred 20 ms after TMF. Whereas on the surface, the findings suggest a near one-to-one dependent relationship between TMF and MRT, it must be noted that, as with MRT, TMF was merely another indicator of the acknowledgement of the target. The fact that maintained fixation usually preceded the manual response may be explained by a saccade-manual response coupling.