Imagining an odour

Mental imagery is proposed to directly engage perceptual pathways to recreate the experience of an item, in the absence of a presented stimulus (Farah, 1988; Kosslyn &

Thompson, 2003; cf. Pylyshyn, 2003). Imagery is closely related to working memory (Tong, 2013), and consequently to directed attention, consciousness, and the ability to actively rehearse or refresh olfactory information in memory (M. R. Johnson, Mitchell, Raye, D’Esposito, & Johnson, 2007; Stevenson, 2009). The re-creation of an olfactory perceptual experience would therefore support an ability to access an internal representation that is analogous to processes in visual memory (Kosslyn, Ganis, &

Thompson, 2001), but findings in this area are equivocal (e.g. Crowder & Schab, 1995;

Rinck, Rouby, & Bensafi, 2009; Royet, Delon-Martin, & Plailly, 2013; Stevenson, 2009; Tomiczek & Stevenson, 2009). However, although investigations of olfactory working memory are limited (see White, 2012, for a discussion of olfactory working memory, and Section 1.3 in this thesis), evidence for such an ability (e.g. Dade, Zatorre, Evans, & Jones-Gotman, 2001; Jönsson et al., 2011; Zelano et al., 2009) suggests some capability to both image and consciously access an internal representation of odours.

This section explores the nature of olfactory imagery, and examines how such an ability might differ to that seen for other modalities.

It is clear that the ability to imagine an odour is poor in comparison to visual, auditory, or even haptic stimuli (Herz, 1996, cited in Herz, 2000; Stevenson & Case, 2005). For example, odour images are self-reported as less vivid and more difficult to produce than other modalities (Ashton & White, 1980). Indeed, some research suggests such an ability may not be possible at all (Crowder & Schab, 1995; Herz, 2000). For example, a mental image is proposed to re-create the perceptual experience of an odour, so paired-associative memory would be expected to show similar performance regardless of whether cues at learning and test were the actual stimulus or a prompt to imagine the stimulus. Herz (2000) showed this was not the case, as cued recall was impaired when the cue switched from an imagine format at study to the actual odour at test, compared to when both study and test were imagine cues. The presence of switch effects for olfactory memory was interpreted as an inability of imagery to reproduce the perceptual experience of an odour. However, it should be noted that there was no alternative modality presented in this task with which to compare these effects of cue-switching.

In contrast, several avenues of research have supported an olfactory imagery ability (for reviews see Rinck et al., 2009; Stevenson, 2009; Stevenson & Case, 2005). Olfactory hallucinations, for example, have been reported in the absence of a sensory stimulus, and is evidence for an ability to recreate the perceptual experience of an odour (Stevenson & Case, 2005). Furthermore, like other modalities, participants imagining the experience of an odour will activate overlapping brain regions with those when actually experiencing an odour (Bensafi, Sobel, & Khan, 2007; Djordjevic, Zatorre, Petrides, Boyle, & Jones-Gotman, 2005; see Kosslyn et al., 2001 for a discussion of the neural basis of imagery for other modalities). Similarly, activation related to odour

processing in the inferior frontal gyrus occurs in anticipation of olfactory task demands, and continues in the short-term memory period beyond removal of the stimulus (Rolls, Grabenhorst, & Margot, 2008).

Tomiczek and Stevenson (2009) demonstrated generation of an odour image through facilitative effects of odour imagery priming in an odour-name association task, though these imagery effects were dependent on the ability of participants to name odours in an earlier task. In three experiments, the authors assessed the effect of olfactory imagery priming for participants classified as good or bad ‘namers’. A key finding was an interaction where d’ scores were selectively improved in the odour imagery priming condition (compared to visual imagery priming and a control condition), and this effect only occurred for the good ‘namer’ group. Importantly, the null effect of visual imagery priming suggests this facilitation was not a semantic effect, as a similar advantage to the olfactory condition would be expected. Instead, the authors suggest the good ‘namers’

have strong odour-name associations, which are reciprocally activated when attempting to imagine an odour. That is, only where an odour-name association is strong will an attempt to imagine an odour produce imagery priming effects. The strong odour-name association is proposed to reciprocally allow the activation of an odour image, and is similar to other demonstrations of a perceptual odour imagery which have shown improved ability after a learned link between odour and its name (Stevenson, Case, &

Mahmut, 2007; Sugiyama, Ayabe-Kanamura, & Kikuchi, 2006).

Together, these findings are suggested to support imagery that includes a sensorial-type representation (Kosslyn, 2003). For example, in their review Rinck et al. (2009) describes this imagery as the consequence of an activated long-term representation (engaged by sniffing when attempting to image the odour), which is subsequently used to generate the sensorial representation. However, it has been argued that these findings

can be accommodated as a capacity for odour imagery which is not available to consciousness (Stevenson, 2009; Stevenson & Attuquayefio, 2013). For example, Tomiczek and Stevenson (2009) propose that their findings supporting sensorial odour imagery may not be specific to a particular odour. Instead, they suggest that generic activation of olfactory neural networks (as a result of activating all extant odour-name associations) produces the priming effect without conscious imagery of a specific odour.

It should also be noted, however, that such an interpretation contrasts the findings in Djordjevic et al., (2005), where an odour detection advantage was shown when the odour image and olfactory stimulus were matched, compared to when participants detected a different odour to the one they were required to image.

Internal representations that are unavailable to consciousness are also considered within the olfactory-centred unitary model described in Section 1.1.3, which suggests that demonstrations of olfactory memory is simply the result of residual activation and decay processes (Wilson & Stevenson, 2006). To be clear, it is suggested that olfaction can demonstrate phenomenal consciousness where there is experience of an olfactory sensation, but conscious access to these contents (through attention, or working memory) may not be possible (Stevenson, 2009). Consequently Stevenson (2009) suggests the effects of priming, hallucinations, and overlapping neural activations may be supported by unconscious imagery, but access to this internal representation for active maintenance or other tasks may not be possible.

However, it should be noted that there are other accounts of consciousness that suggest dissociation between phenomenal and access consciousness is not appropriate.

Specifically, they suggest consciousness should instead be considered as a hierarchy of access to featural and semantic information (Kouider, de Gardelle, Sackur, & Dupoux, 2010). This alternative model describes levels on a hierarchy that are accessed

independently, and thus allows a graded form of access consciousness. Graded access to consciousness may explain individual differences in olfactory imagery, and indeed a continuous scale of olfactory imagery ability has been proposed based on participant expertise (Arshamian & Larsson, 2014). Olfactory imagery is more vivid in olfactory experts than non-experts (Gilbert, Crouch, & Kemp, 1998), and expertise is associated with imagery that is consciously accessible (Plailly, Delon-Martin, & Royet, 2012;

Royet et al., 2013; Stevenson & Attuquayefio, 2013). Olfactory imagery capacity is drastically affected by individual differences (Arshamian & Larsson, 2014), and may be related to semantic knowledge (Stevenson et al., 2007; Tomiczek & Stevenson, 2009) or perceptual experience (Delon-Martin, Plailly, Fonlupt, Veyrac, & Royet, 2013; Plailly et al., 2012). In Delon-Martin et al. (2013), for example, structural reorganisation of olfactory brain regions related to imagery was observed for those with extensive olfactory experience (perfumers). However, expertise is not necessarily essential in imagery. For example, other findings have shown an advantage for self-reported olfactory imagers in a same-different memory task was unrelated to the ability to identify odours (Köster et al., 2014). Taken together, there is some support for an ability to consciously access a perceptually-based olfactory representation, though expertise may be necessary for imagery and related working memory functions to occur that are analogous to other modalities.