Abstract
Attention is the biased selection of information for representation in a limited capacity working memory system. It is supported by top-down (goal-driven) and bottom-up (stimulus-driven) selection, and largely determines what information is encoded into memory. However, the role of attention mode in memory is yet to be clarified. The purpose of this thesis is to better understand how these two modes of attention might differentially impact memory processing, using a combination of behavioural, eye tracking, and functional magnetic resonance imaging methods.
A new task is introduced that manipulates goal- and stimulus-driven attention during memory encoding. This demonstrated that, when attention is captured (bottom-up attention) by a salient target during encoding, responses were faster and more accurate compared with when targets were identified through visual search (top-down attention). Bottom-up attentional capture was tentatively linked to a pupillary orienting response in an eye tracking study and was strongly associated with activation of the ventral attention network in an fMRI study. Conversely, top-down attention search was associated with more gaze time on distractor stimuli, sustained increases in pupil dilation, and activation of the dorsal attention network. Crucially, despite clear effects during memory encoding, across all experiments using this task, bottom-up attentional capture by the salient target did not produce a meaningfully higher rate of later recognition memory, and modelled estimates of recollection and familiarity also did not appear to differ.
In a final study, the classic spatial cueing task was adapted to explore whether task-irrelevant distractor stimuli are better encoded when they are salient. Top-down attention was directed to target items and away from distractors. Rare salient distractors were presented to test whether there are memory differences due to bottom-up attention (via salience processing). Again, despite clear effects of spatial cueing, and an indication that salience also exerted an effect on performance, there was no evidence that salience led to better later memory for distractors, nor that it was sufficiently disruptive to target encoding.
Based on the data presented in this thesis, it appears that top-down and bottom-up attention exert their effects primarily on momentary behaviour in the service of task execution, but that their effects on later memory performance may depend on other factors. Top-down attention may facilitate memory encoding to the extent that accomplishing task goals is supportive of encoding processes, such as selecting task-relevant items or engaging in elaborative processing. Bottom-up attention modulates selection priority, but its effects on later memory could occur indirectly: intention to encode and the subjective value associated with processing salient items may interact with this mode of attention to produce differences in memory. This means that top-down and bottom-up attention may affect memory only indirectly and via intermediary processes.