The 2-back task (Figure 2) presents participants with a sequence of stimuli, where each item must be compared with the stimulus presented 2 trials previously, whilst
3.5 General Discussion
The goal of the current chapter was to apply the normative data produced in Chapter 2 to replicate the previous finding of a working memory advantage for odours that are verbalisable (Jönsson et al., 2011), and to explore this advantage in relation to strategies employed in the n-back task.
The observed advantage in the n-back task for verbalisable odours found across Experiments 1-4 serves to validate the normative data collected in Chapter 2. The normative data were used to separate odour sets on verbalisability in all of the above experiments. The replication of a verbalisability advantage (Jönsson et al., 2011) supports conclusions that verbal labelling, or some covariate to verbal labelling ability, facilitates working memory performance in the n-back task. Although the interaction in Experiment 1 lent support to the proposal that verbal labels are generated and refined throughout the task, this effect was not replicated in future experiments. Furthermore, whilst the advantage for verbalisable odours was robust throughout each replication of the task as measured by A’ sensitivity, some variation across experiments were observed as to whether the effect was driven by a change in hit rate (Experiment 3 and 4), false alarms (Experiment 2), or both (Experiment 1).
Demonstration of olfactory working memory in the n-back task (with above chance sensitivity for low verbalisability odours) demonstrates that an odour representation can be re-activated and compared to the presented stimulus. However, n-back performance for both low and high verbalisability odours may be mediated by the ability to verbally re-code the items that subsequently allows rehearsal (Murphy et al., 1991). Furthermore, if verbal rehearsal is not occurring, it is debated whether such representations are consciously accessible and thus available to active maintenance and updating processes required in the n-back task (Arshamian & Larsson, 2014; Stevenson, 2009). Experiment
2 provides evidence that performance in the n-back task is not reliant on a verbal rehearsal process for odours either high or low on normative verbalisability.
Specifically, reducing verbal rehearsal opportunities (through inter-trial CA) did not attenuate the memory advantage for the verbalisable odours, nor produce a general decline in performance. In addition, Experiment 2 showed some effect of concurrent rotation on false alarm rates, potentially suggesting that the retention interval was used to maintain stimuli through a refreshing process (Raye et al., 2007). These findings warrant further investigation to assess the executive resources utilised for maintaining and updating olfactory information in the task. The concurrent articulation findings however are consistent with suggestions that olfactory memory is not dependent on verbal working memory processes (Andrade & Donaldson, 2007). These findings expand upon previous results by demonstrating a lack of verbal rehearsal during a task that ostensibly has both maintenance and updating requirements.
However, the n-back task, like other working memory tasks, can be influenced by automatic processes that assess the familiarity of items to make a task judgement (Loaiza et al. 2015). Indeed, an n-back decision may be made based on a familiarity signal and thus have no requirement for rehearsal in working memory, either through acceptance or rejection based on a strength criterion or through a low-control process where familiarity is compared to a signal-strength estimate for a target item (Juvina &
Taatgen, 2007). In Experiment 3, the reliance on familiarity-based judgements or controlled strategies (recollection-based judgements) for completing the n-back task were assessed across odour verbalisability. The remember-know paradigm revealed a reduction in item recollection for low verbalisability odours with no corresponding increase in familiarity-based responding. This suggests that the advantage for high verbalisability odours is related to more successful application of a control strategy, and
not a qualitative shift away from a control strategy towards either a low-control assessment of familiarity, or a criterion-based assessment of familiarity for low verbalisability odours. Instead, a large proportion of correct responses were the result of guess responses for low verbalisable odours. It was suggested that this is the result of reduced conceptual saliency, which affects the ability of participants to maintain and update bindings between the odour and its context in the task.
Finally, in Experiment 4 it was assessed whether the observed working memory advantage for high verbalisable odours was mediated by perceptual familiarity to the odours; based on previous studies relating high familiarity to fewer cognitive resources at encoding (Reder et al., 2015) and increased discriminability (Wilson & Stevenson, 2003a). Multiple presentations of odours in a preliminary task were applied to artificially induce item familiarity to previously unfamiliar odours. However, the experiment found no evidence to support the proposal that mere exposure can improve olfactory working memory, though the findings supported a disruptive effect of familiarisation, perhaps as a result of an unreliable familiarity signal. Consequently, although Experiment 4 did not reveal the expected findings related to perceptual familiarity, it does support proposals that item familiarity is important in an n-back task decision, and that the verbalisability advantage for odours occurs from something other than perceptual familiarity (e.g. conceptual saliency).
In summary, this chapter has replicated a verbalisability advantage for odours in the n-back task, but has presented evidence against this advantage being due to verbal rehearsal (Experiment 2), or perceptual familiarity (Experiment 4). The performance advantage instead appears to be due to an increase in the ability to maintain and update bindings between the odour and its context in working memory (Oberauer, 2005), though it should be noted that there was no evidence to support a shift in strategy
(Experiment 3). Indeed there was a general increase in false alarms in a concurrent rotation task that supports similar application of working memory resources in both verbalisability conditions (Experiment 2).
Chapter 4: An individual differences analysis of cross-modal n-back ability