Quantitative Visuospatial Memory Tasks

To overcome the influence or potential use of executive attentional resources within a working memory task, it may be possible to use a different task protocol. In this type of task, known as a change detection task, participants are shown a brief display of an encoding image, followed by a retention interval which is displayed before a final retrieval or proble image is presented. The retrieval or probe image can display either a change or no change of which has to be identified by the participant. Change detection tasks have conventionally been used to investigate working memory capacity in terms of information load (Alvarez and Cavanagh, 2004), looking at visual working memory capacity in terms of information load (Alvarez & Cavanagh, 2004; Phillips & Christie, 1977), looking at the working memory of faces (Scolari, Vogel and Awh, 2008) and have also been used in developmental contexts to investigate working memory capacity in young children (Cowan et al., 2006; Riggs, McTaggart, Simpson and Freeman, 2006; see also Hamilton et al., 2013).

An early change detection paradigm, constructed by Luck and Vogel (1997), as seen in figure 2.3., has been widely used throughout the visual working memory literature. Luck and Vogel

36 (1997) used their initial paradigm with college students, however, since then the paradigm has also been successfully used with children (Cowan et al., 2005)

In their 1997 work, Luck and Vogel showed participants arrays of 1, 2, 3, 4, 8 or 12 coloured squares. A retrieval array was shown, for 2000 milliseconds, consisting of the same number of squares, and participants had to decide if there was a change (or not) in the cued square in each retrieval array. Initially the encoding arrays were displayed for 100 milliseconds with a 900 millisecond maintenance interval; however, a comparison was made to the encoding display of 500 milliseconds to see if the longer encoding time encouraged any encoding differences. The authors identified that there were no differences in performance of the change detection task when participants were exposed to both the 100 millisecond and 500 millisecond encoding array conditions. The encoding time of 500 milliseconds has subsequently been used throughout the change detection literature (e.g. Alvarez and Cavanagh, 2004; Luck and Vogel, 1997; Riggs et al., 2006; Scolari et al., 2008).

Figure 2.3. An example of a change detection paradigm (Luck and Vogel, 2013). This was successfully used by Luck and Vogel (1997) and also Riggs et al. (2006). Participants have to

decide if the second array is the same or different as the first array.

37 In a more recent review, Luck and Vogel (2013) defined visual working memory tasks as

‘First, to qualify as VWM, it is not sufficient that the information was acquired through the

visual modality; the representation of the information must be visual in nature. If the observer stores a verbal or amodal conceptual representation of the sensory input, we no longer consider it to be a visual memory’(Luck and Vogel, 2013, page 2).

Here, it was suggested that visual working memory tasks should use visual representations only and should have no verbal influences such as those that have been shown with the VPT (Brown et al., 2006; Brown & Wesley, 2013).

To control for verbal influences in their original work, Luck and Vogel (1997) used a verbal load task to eliminate any use of verbal representations. Participants were asked to hold two digits in their mind and recall them at the end of each trial, essentially preventing the use of verbal material during the visual working memory task. This inhibited any verbal influences, such as those seen in the VPT and the Corsi Blocks Task (Thompson et al., 2006).

As a result of this ‘visual only’ perspective, the current doctoral thesis will aim to use a change detection paradigm as one of the aims of the thesis is to specifically look at visual working memory tasks. The Corsi Blocks Task and the Visual Patterns Test will not be used due to the increasing research regarding the use of potential executive attentional resources during these tasks (Brown and Wesley, 2013; Hamilton et al., 2003; Rudkin et al., 2007).

Although current researchers have suggested change detection protocols to be an appropriate source of visual working memory assessment, there are criticisms of these tasks. As previously discussed, the possibility of proactive interference within the task is one which cannot be identified fully. Hartshorne (2008) used a 1000 millisecond encoding interval in their work, presenting effects of proactive interference and potential semantic representation use with this extended encoding duration. Phillips and Christie (1997) also used a 1 second

38 encoding interval, demonstrating that this encoding time was long enough to enable to use of long term memory. However, Lin and Luck (2012) used an alternative encoding duration of 100 milliseconds to show no proactive interference effects within their colour change detection paradigm. This shorter maintenance interval was show to have no interference effects with different shapes, circles and orientated bars, concluding that the shorter maintenance interval was more appropriate to use in a working memory paradigm. As Luck and Vogel (1997) had suggested no difference between the 100 millisecond and 500 millisecond maintenance interval in their work, the current thesis will employ a 500 millisecond maintenance interval as this has been used in both adult (Fukuda et al., 2010) and child contexts (Riggs et al., 2006).

Another point to consider is the Hierarchical Encoding perspective offered from Brady and Alvarez (2011, 2015) which suggests that there are organisational and long term memory influences during full arrays probe contexts. Brady and Alvarez suggested that when an encoding array of multiple probes is presented, the participants unconsciously compute

‘Ensemble Statistics’ with a mean amount of the size or colour of all items in an array as well as the initial information about each individual items in the array. Brady and Alvarez suggested that this second form of representation is susceptible to long term memory organisational influences. Because of this, there could be influences of long term memory within change detection tasks where items are recalled as full arrays and not individual stimuli as previously proposed by Luck and Vogel (1997). This is one point to consider with regards to visual change detection tasks as it could potentially mean that arrays larger than 1 may still be susceptible to long term memory influences even when the task has been designed to eliminate the use of verbal coding.

Other criticisms of the change detection protocol retrieval contexts have been made. Wheeler and Treisman (2002) conducted investigations using Luck and Vogel’s change detection task,

39 giving details of the use of both single and full retrieval probes. Wheler and Treisman (2002) suggested that the use of a single retrieval probe would eliminate any need to bind information within memory, such as colour and location; therefore, memory errors would be reduced. Multiple retrieval probes may increase the task demands and may potentially need greater attentional control of the stimuli presented. Brady and Alvarez (2011) contrasted this, suggesting that a full array can be seen as an advantage to the participant. In their investigation, a full array was presented to participants and researchers demonstrated that participants used the full array to form the representation of the item to be recalled. The issue of single versus multiple retrieval probes is one of which will be investigated during the next chapter of the current thesis. Before implementing any change detection task, researchers will pilot the change detection task using a single and multiple retrieval probes to discover if any differences in performance are present.

In summary, the change detection tasks discussed above, have all been used in contexts to assess ‘large’ or quite discrete changes in visual memory and consequently are known as quantitative memory tasks. For example, Luck and Vogel (1997) could have changed a red square to a blue square. These tasks relate to the previously discussed Discrete Slot Model (Luck and Vogel, 1997) in chapter 1, which focus upon the number aspect of visual working memory capacity. In more recent research (e.g. Bae and Flombaum, 2013; Dean, Dewhurst &

Whittaker, 2008; Dent, 2010; Thompson et al., 2006), small or continuous changes in visual memory have been assessed using a different protocol of change detection tasks. These tasks may be labelled as qualitative tasks as they do not measure capacity of visual working memory, but the fidelity, resolution or quality of the maintained visual representation. The current doctoral thesis will aim to look at both quantitative and qualitative representations in visual working memory. These tasks are associated with the Shared Resource Model (Bays et

40 al., 2009) which is concerned with looking at the fine grained representations within working memory instead of the amount of objects that can be stored.