Working Memory and Visual Search Efficiency

Many theories have implicated working memory involvement during visual attention and particularly during a visual search task (Awe et al., 2006; Cowan, 1995; Desimone

& Duncan, 1995; Duncan & Humphreys, 1989). The majority of these claim that visual attention is influenced by object representations stored in working memory which can be used as either reference points or a guide to the most relevant information. However, Woodman, Vogel, and Luck (2001) proposed that the constant transfer of such information into working memory was a highly inefficient process and that another mechanism must be implicated in visual search efficiency. They investigated the role of working memory in visual attention by using a dual-task paradigm. Participants were presented with four, eight, or twelve black objects (outline of squares), in groups of four. The target object had a gap on either the top or the bottom, while the nontargets had a gap on either the left or right side. The memory array consisted of four coloured squares placed around a central fixation cross. In the dual-task condition, participants were initially presented with a memory array, followed by a visual search task, and then a memory-test display. Participants were asked to respond whether the memory test display was identical to the memory array that was initially presented, or whether one of the squares was a different colour. In the single-task condition, only the visual search task was administered.

Woodman et al. (2001) revealed that participants were able to perform both tasks without a significant disruption to either such that the search slopes were virtually identical in both conditions. However, they also found that a constant value was added to the visual search reaction times when participants maintained information in working memory, such that there appeared to be a general slowing that either preceded or followed the search task. They speculated that this was related to a delay in the onset of

search processes or a delay in post-search processes such as response selection (see Jolicoeur & Dell’Acqua, 1999). While accuracy during the search task was very high, accuracy in the memory task was reduced particularly in the dual-task condition, which suggested that memory capacity was exceeded. Woodman and associates suggested that information stored in visual working memory did not disrupt visual search, even when visual working memory was filled to capacity. In contrast to other theories, these results suggested that during visual search, targets were not stored in working memory and that objects could be attended to purely at a perceptual level without being automatically entered into working memory (Woodman et al).

These findings demonstrate that working memory load influences the ability to complete an additional task both efficiently and accurately. While some theorists believe that working memory content assists visual search (Desimone & Duncan, 1995), others claim that increased working memory load reduces accuracy (Woodman et al., 2001).

Oh and Kim (2004) examined the interactions between visual selective attention and visual working memory using the dual-task paradigm developed by Woodman et al. (2001). They compared selective attention with both a spatial (Experiment 1) and nonspatial (Experiment 2) working memory task during visual search. Participants completed a change-detection task which required them to determine whether there was a change to the location or colour of the memory test probe, from the initial memory array. Interestingly, they found that distractor interference varied based on the nature of

the working memory task. In the spatial working memory task, visual search performance was affected by the spatial working memory task, and spatial working memory load impaired the visual search process. Oh and Kim suggested that the same limited-capacity mechanisms appeared to be shared by both the visual search process and spatial working memory storage. In contrast, the nonspatial working memory task did not appear to interfere with visual search efficiency and the search process did not interfere with the maintenance of nonspatial information in working memory. Taken together, these results suggested that spatial and nonspatial working memory loads interacted differently with the visual search process.

In a follow-up study Woodman and Luck (2004) also examined the possibility that maintaining spatial information in working memory would interfere with visual search processes. They also used an experimental procedure similar to Woodman et al. (2001) in which they slightly modified the working memory task but kept the visual search task identical. For the working memory task, they used a location change detection task which required participants to remember the spatial locations of two objects. Each trial began with the sequential presentation of two white dots which participants were asked to remember and was followed by a test array in which both dots were presented simultaneously. In the dual-task condition, participants were provided with a search task (described previously; Woodman et al) during the retention interval of the memory task. They found that remembering a relatively modest number of spatial locations affected the efficiency of the search process. Consistent with Oh and Kim (2004), they found that as set size in the search task increased, memory accuracy became

progressively more impaired and vice versa. Woodman and Luck proposed that maintaining spatial representations in working memory may interfere with visual search which supported the hypothesis for the existence of separate working memory systems. In addition, these separate working memory systems may be linked to the dorsal and ventral pathways in the posterior visual cortex which are specialized in processing ‘where’ and ‘what’ information, respectively (Woodman & Luck).

A number of studies have suggested that visual search efficiency is determined by the nature of the working memory task (D’Esposito & Postle, 1998; Han & Kim; 2004; Owen, Evans, & Petrides, 1996; Petrides, 1989). Previous experimental designs have used dual-task paradigms in which participants were asked to complete a visual attention task in addition to a memory task with either a recognition format (Lavie et al., 2004) or a memory test array (Oh & Kim, 2004). However, these studies typically examined the temporary storage of information. In contrast, Han and Kim (2004) argued that working memory encompassed more than just the temporary storage of information (i.e. object representation). They suggested that working memory included a number of different processes such as multiple-task coordination, distractor interference prevention, memory updating, set-shifting, and manipulation/transposition of information (see also D’Esposito Postle, Ballard, & Lease, 1999).

In their series of studies, Han and Kim (2004) investigated the difference between the ability to purely maintain information temporarily in working memory and the ability to manipulate it in some way. They also used the same dual-task paradigm employed by

Woodman et al. (2001). During the maintenance condition, participants were required to remember either a number or a letter, while during the manipulation condition, they were either asked to count backwards by three’s from a specified number or to reorder letters into alphabetical sequence. Performance in the single-task condition was compared to performance in the dual-task condition. Results showed that the ‘manipulation’ working memory task produced significantly steeper search slopes in the dual-task condition compared to the single-task condition. This suggested that performing a working memory task which required the use of higher-order processes impaired efficiency of visual search. In contrast, the simple ‘maintenance’ working memory task did not appear to affect visual search efficiency; the search slopes in both the single- and dual-task conditions were nearly identical. These results indicated that visual search efficiency was not impaired when simple maintenance of information in verbal working memory was required. Thus, Han and Kim proposed that while higher- order processes are implicated in working memory tasks which require the manipulation of information, these same processes may be inactive when simple maintenance of information is only required. These studies highlight that the nature of the working memory task may determine the extent of efficient visual attention and distractor interference.