This study applies an individual-differences approach to further examine whether available odour semantic information (through normative odour verbalisability) is related to the utilisation of controlled working memory resources. The experiment assesses the relationships between n-back performance using low verbalisability odours, high verbalisability odours, abstract shapes, and letter stimuli. Furthermore, individual differences in discriminability are also considered, as discriminability is a limiting factor in any measure of working memory performance (Jönsson et al., 2011).
A relationship is expected between visual and verbal n-back performance due to previously observed similarities in working memory performance across these modalities (e.g. Schmiedek et al., 2014). However, the prediction regarding relationships with odour back tasks are less clear. If performance in these multiple n-back tasks arises from a common working memory resource, the tasks should covary.
The dual-task findings in Experiment 3.2 suggest that similar processes may be engaged for low verbalisability and verbalisable odours (though the evidence for this, and subsequent evidence from the remember-know task are equivocal on this), and would therefore predict this relationship between all n-back task modalities despite lower working memory performance for the low verbalisability odours. However, if the availability of semantic information mediates application of working memory control
processes, a relationship between verbalisable odour working memory and the verbal and visual tasks is predicted, but with no relationship to low verbalisability odour memory. Finally, it is predicted that performance on both low and high verbalisability discrimination tasks will be related to olfactory performance in both low and high verbalisability n-back tasks, due to the perceptual processing requirements proposed for all versions of the n-back task.
4.1.1 Method 4.1.1.1 Participants
Fifty-six participants (44 females, 12 males, mean age = 23.91, SD = 6.64) were recruited from Bournemouth University as part of a course credit requirement. The same exclusion criteria from earlier chapters were applied. Ethical approval was gained for all aspects of the study through the Bournemouth University ethical procedures.
4.1.1.2 Materials
Olfactory stimuli. Fourteen olfactory stimuli were selected for use in this experiment (Table 3). Half of these were classified as low verbalisability, and half as high verbalisability, based on normative ratings in Chapter 2. Odour sets differed significantly on these verbalisability scores, t(12) = 12.96, p < .001, d = 6.93, BF10 >
1,000, and on familiarity scores, t(12) = 22.22, p < .001, d = 11.88, BF10 > 1,000.
Furthermore, the odour sets were balanced on ratings of intensity, and Bayes Factor analysis revealed support for no difference in intensity, t(12) = -0.19, p = .851, BF01 = 2.21. Whilst pleasantness ratings did differ between the two sets, t(12) = 8.00, p < .001, d = 4.28, BF10 > 1,000, there was evidence for no difference in the hedonic strength across the two sets. This was calculated as the deviance from a neutral midpoint on the pleasantness rating scale (Chapter 2), t(12) = 0.67, p = .515, BF01 = 1.93. Responses
were collected using a Cedrus Response Box, and recorded using Superlab 5 (Cedrus, 2015).
Table 3
Normative ratings and grouping of olfactory stimuli used in the low and high verbalisability odour n-back tasks.
Odour Task Verb. Fam. Int. Pleas. Hed. Str.
Lime* High 2.73 5.70 5.06 5.16 1.40
Pear High 2.62 5.82 5.16 4.40 1.40
Blackcurrant High 2.44 5.67 4.85 5.48 1.73
Marzipan High 2.73 6.12 5.27 4.96 1.65
Spearmint High 2.71 5.90 4.96 5.08 1.48
Aniseed Balls High 2.61 5.88 5.40 3.98 1.50
Sports Rub High 2.69 5.60 5.52 4.21 1.08
Cheddar Cheese* Low 1.24 3.14 5.27 2.35 1.86
Ginger Low 1.66 3.39 5.22 3.10 1.39
Sea Shore Low 1.53 2.96 5.20 2.20 1.84
Rum Barrel Low 1.26 3.10 5.18 2.68 1.56
Carbolic Soap Low 1.05 3.61 5.10 3.12 1.33
Patchouli Low 1.62 3.55 5.06 3.02 1.31
Mouse Low 1.53 3.36 5.06 2.70 1.50
* Buffer items not included in analysis
Visual stimuli. Seven irregularly-shaped polygons (Chuah, Maybery, & Fox, 2004) designed to prevent verbal rehearsal strategies (Attneave, Arnoult, & Attneave, 1956;
Smith et al., 1995) were used as 2-back stimuli. These were presented using
Opensesame (Mathôt et al., 2012) in the centre of a 22-inch 60hz monitor as black line drawings on a white background, within a black border square of 62px by 62px (See Figure 10).
Verbal stimuli. Eight phonologically dissimilar consonants were selected (B, F, H, K, M, Q, R, X) (Kane et al., 2007), and displayed centrally on a 22-inch 60hz monitor in size 21pt. monospaced font. Stimuli were randomly presented in lower or upper case to limit responses based on the visual features of the letters (although one might argue that f, k, and m are visually similar in lower and upper case forms), and stimulus presentation timings and trial responses (Figure 10) were controlled by OpenSesame.
4.1.1.3 Design
Olfactory 2-back task. A continuous yes/no recognition paradigm was employed on two 26-item sequences, where each trial necessitated a judgement as to whether the item had been presented two items previous. The low and high verbalisability odours were presented in the same blocked design used in Experiments 3.1 and 3.2, and Jönsson et al. (2011, Experiment 2), meaning participants experienced 26 trials of one odour set followed by 26 trials of the other set. A unique low or high verbalisability odour, corresponding to the odour set being tested, was presented as a buffer item in the first two trials of each 2-back task, where a ‘no’ response was guaranteed. Each remaining odour appeared as a lure three times and as a target once. In addition, the presence of close-lures was increased compared to earlier experiments to more closely match the number of targets, with a task containing 5, or 6, n+1 and n-1 lures. This adjustment was important for two reasons. First, it discourages a reliance on a familiarity-based strategy simply because the payoff from using a recollection-based strategy was not worth the use of additional resources (Ralph, 2014). Furthermore, it allows analysis of responses based on only recent-lure and target decisions, meaning judgements that may be based
solely on a familiarity criterion (non-recent lures) are not included in the index of working memory ability.
A blocked design allowed the two sets to be considered as independent memory tasks, minimising cross-contamination of items from the other odour set. That is, a criticism of Experiments 3.3 and 3.4 was that the odour presented in the n-1 position, if different across the dimension of interest to the 2-back item and probe, may have influenced the probe’s acceptance or rejection through some combination of familiarity-based responding and a process-of-elimination strategy.
The orientation of targets and lures within tasks were identical for all participants, but were counterbalanced such that the half of the participants performed the low verbalisability task first, and the other half the high verbalisability task first. In addition, order of trials (targets and lures) was counterbalanced between the low and high verbalisability tasks. The former counterbalancing accounted for practice effects, whereas the latter for differences in trial order difficulty. That is, the slightly different number of close lures in a sequence may have affected the difficulty of the task, but this was balanced across tasks between participants.
Visual 2-back task. The visual 2-back task consisted of 2 blocks of 26 items, in an identical trial sequence to the two olfactory sequences. Two identical buffer images preceded the 24 critical 2-back trials in each block, and did not occur again in either sequence. The presentation order of the two visual blocks was randomised across participants.
Verbal 3-back task. Pilot work (data unavailable) suggested that a 2-back task with verbal stimuli was close to ceiling, meaning a 3-back task was instead selected for verbal stimuli. The task necessitated a judgement whether the currently presented item
matched the item presented three trials previous. Consequently, the first three items in a block sequence were guaranteed ‘no’ responses, after which the fourth trial was compared to the first, fifth to the second, etc. Participants completed 3 unique blocks of 40 trials, and were given the opportunity for a break between each block. For each sequence, a letter appeared once as a target, and four times as a lure, totalling 8 targets and 32 lures in each block. Furthermore, a minimum of 7 n+1, n-1, and n-2 lures occurred within a sequence, and a maximum of 10. The trial order was pre-generated and the same for all participants. It should be noted that this task therefore differs substantially to the 2-back tasks, particularly in terms of general working memory load and the presence of recent-lure trials.
Figure 10. Schematic diagram of verbal 3-back and visual 2-back tasks.
Olfactory discriminability task. A paired discrimination task was employed where participants made a same/different judgement for two odours presented in succession.
There were 42 possible non-match combinations, where an odour within a set was paired with every other odour in that set (i.e. a high verbalisability odour was only tested against other high verbalisability odours). There were 14 targets, with each odour