movement characteristics. Those relating to the feedback manipulation are addressed first, before moving on to the overall pattern of results as they relate to each training group.
It was hypothesised that manipulating feedback would give rise to either a decrease in target processing time for the MT group (when feedback was not available) or a increase (when feedback was available). The first prediction was attached to the idea that in the absence of feedback participants would become more entrenched in following the predictable sequence of target presentation, and that as a result the reciprocal relationship between the Move and Fixate centres would lead to the target being inspected for insufficient time. An accuracy decrease was therefore proposed in connection with this potential outcome. The second prediction was based on the argument that providing feedback would increase stress about responding correctly; the reciprocal relationship between the Move and Fixate centres could therefore lead to the opposite outcome: an increase in target
accuracy, because no accuracy decrement was found in the previous version of this experiment when feedback was available (Exp 3b), an accuracy decrease was not linked to this prediction.
The latter account was supported in part by the data, with the MT group showing increased target processing time (revealed in the gaze duration measure). However, the effect was not feedback specific, occurring irrespective of whether feedback was presented or not. As such, there was no evidence for a reduction in target processing time, and associated decrease in accuracy in the MT group. On the surface therefore it appears that MT does increase target processing time and that the feedback manipulation was unnecessary to reveal this effect. (Although note that feedback was not entirely insignificant in affecting oculomotor behaviour: there was a marginal improvement for the MT group in terms of fewer fixations per trial when feedback was withdrawn, perhaps indicating greater adherence to the training sequence as suggested).
Not only does the MT group show an increase in target processing time at the end of a trial, possibly due to competition between completing the response discrimination and suppression of the next eye movement in the sequence, but the BT group, despite utilising the same training sequence of target presentation, do not show such an increase. Conjoined training seems to offset the artificial extension in the time needed to distinguish the correct response once focussed on the target, seen when directed with MT in isolation. This result, combined with the overall pattern in the eye movement and manual response data, suggests that performance is best when training is directed at both the Move and Fixate centres.
As individual fixation durations on the target were not significantly different between groups (revealed in the mean target fixation duration measure) one must conclude that the acquisition of information from each target fixation of the MT
group is less than optimal, a drawback remedied by providing Fixate training in concert with Move training. The BT group require fewer fixations (of comparable duration to the MT group) to process the target fully. This is not to say that the BT group apprehend the same information as the MT group with greater economy (the target’s colour and shape are irrelevant to the BT group); only that attention to redundant detail is removed by the provision of Fixate training. Although Findlay & Walker’s (1999) framework primarily deals with single fixations, the present results suggest that when the task relies so heavily on top down control, effects on the processing efficiency of each fixation may be evident only in aggregated oculomotor statistics not individual fixation durations per se.
I mentioned at the beginning of this chapter that incorporating eye tracking into the present paradigm would also shed light on more classical conceptions of top down control (i.e. performing the task when untrained –naïve to the training
contingencies –still depends considerably on high level cognitive influences). Our understanding of visual search and feature integration originates from a literature which draws heavily on notions of covert attention, but less so on eye movements; so called “overt attention” (Duncan & Humphreys, 1989; Quinlan, 2003; Shen & Pare, 2006; Treisman & Gormican, 1988; though it is noted that big steps have been made to bridge this gap in recent years e.g. C. C. Williams & Pollatsek, 2007; D. E. Williams, Reingold, Moscovitch, & Behrmann, 1997). The eye movement data from the present study add to the advancements being made in understanding how attention
translates into eye movements in visual search, revealing that, despite the difficulty of the task, participants can still move their eyes to the target then interpret the response relatively quickly, even when they are unaware of the training strategies (it is noted that upon debrief none of the participants spontaneously noticed any of the training contingencies). Moreover, performance improved in the NT group despite
the fact they were given no strategic advice: their overall visual search performance improved marginally, but also their ability to carry out the response discrimination became much better as the experiment progressed – gaze durations on the target actually decreased. This indicates that, despite the time pressure imposed (each trial = 4000ms, maximum), the NT group learn to classify the feature conjunction
response criterions more resourcefully. Indeed, upon debrief, some participants from the NT group expressed that they developed their own strategies to help them respond correctly. For example, they may remember certain sub-array configurations and the appropriate response.
The apparent improvement of the NT group with the response discrimination leads on to another point concerning the extended gaze durations of the MT group; do the increased gaze durations seen when training is directed at the Move centre in isolation really reflect a competitive interaction between processing the target and making the next eye movement in the sequence? Perhaps the explanation is less esoteric: the NT group have a smaller portion of the trial remaining once they have located the target, by virtue of necessity therefore they learn to interpret the response more efficiently; conversely, because the visual search requirement is effectively removed for the MT group, they might spend the remainder of the trial trying to work out the correct response, and do not improve with this aspect of the task because the available time is not constricted. This argument detracts from the proposal above that the MT group’s increased gaze durations are due to reciprocal antagonism between the Move and Fixate centres. It is worth reiterating however that whatever the reason for the apparent extension in target processing time seen in the MT group, this effect is eliminated when directed in both Move and Fixate training (i.e. as evidenced by the BT group).
If, in the interest of parsimony, one favours the alternative reason for the MT group’s increased target processing time, should we accept that when training is directed solely at the Move centre it does not hinder information processing sub- served by the Fixate centre? On the basis of this evidence alone we should not. A plausible reason it was difficult to find solid evidence of competitive inhibition between the Move and Fixate centres in Experiment 4 is because such competition is less likely to arise unless the next eye movement in the MT sequence competes with target processing within the same trial. With the present paradigm participants do not need to move their eyes to the next target location until the trial is over, therefore lessening the source of conflict between target processing and eye movement execution (sub-served by the Move and Fixate centres respectively). The next series of experiments will address this issue, placing spatially separated items in a visual array in direct conflict intra-, rather than inter-, trial.