The aim of the present study was to investigate the subjective impact of visual imagery absence on cognition. To achieve this, we compared self-reports of aphantasic individuals with those of general population individuals (with self-reported intact visual imagery) on several cognitive domains including multi-sensory imagery, episodic memory, trauma response, dreaming and daydreaming, and spatial abilities. The main results sections presented here all describe between-group tests comparing our aphantasic sample with our first control group of age-matched participants recruited from MTurk (see Tables S2–6 in Supplementary Information). For replication comparisons with our second control group sample of undergraduates, see section at end of Results titled “Control Group 2: Replication Analysis”.
Control Group 1: Main Comparisons
We first examined group differences in visual imagery vividness. As expected based on previous findings9,10, aphantasic participants rated their visual imagery ability on the VVIQ as being significantly lower (17.94 ± 0.223, with many (70%) scoring at floor, i.e. 16) compared to control group 1 (58.12 ± 0.888; Mann-Whitney U = 427.5, p < 0.0002, r = 0.87, BF10 = 1.41e12, 2-tailed; see Fig. 1 red section and Figure S1 in Supplementary Information; Fig. 1 depicts median-centered data with the aphantasia group denoted by red plots and control group 1 by blue plots throughout; Figures S1–5 show raw scale scores and distributions). This self-reported qualitative absence of visual imagery vividness was mirrored by significantly lower scores than controls on the object imagery component of the OSIQ (Mann-Whitney U = 372, p < 0.0002, r = 0.85, BF10 = 446,931.23, 2-tailed; see Fig. 1 red section and Fig. S1), which measures the perceived ability to use imagery as a cognitive tool in task-relevant scenarios. Our data also showed that individuals with aphantasia not only report being unable to visualise, but also report comparatively reduced imagery, on average, in all other sensory modalities (measured using the QMI), including auditory (U = 6,152, BF10 = 5.01e11), tactile (U = 4,473, BF10 = 4.90e9), kinesthetic (U = 5,151, BF10 = 1.04e11), taste (U = 3,069.5, BF10 = 4.82e26), olfactory (U = 3,439.5, BF10 = 2.73e9) and emotion (U = 6,670.5, BF10 = 4.81e12) domains (all Mann-Whitney U-tests, p < 0.0002, r = 0.65–.78, 2-tailed; see Fig. 2a and Fig. S1). It is noteworthy, however, that despite reporting a near total absence of visual imagery on the QMI (Mann-Whitney U = 620.5, p < 0.0002, r = 0.87, BF10 = 1.07e9, 2-tailed; see Fig. 2a) and significantly lower total QMI scores overall compared to controls (Mann-Whitney U = 1,868.5, p < 0.0002, r = 0.79, BF10 = 6.47e12, 2-tailed; see Fig. 1 red section, second panel from top), only 26.22% of aphantasic participants reported a complete lack of multi-sensory imagery altogether (rating each question in each QMI domain as “1: No sensory experience at all”). The remainder of our aphantasic sample (73.78%) reported some degree of imagery in non-visual sensory modalities (albeit significantly reduced compared to controls; see Fig. 1 red section, and Fig. 2a), suggesting potential sub-categories of aphantasia.
Aphantasic individuals described a significantly lower ability to remember specific life events in general (Event Memory component of the SAM; Mann-Whitney U = 8,865, p < 0.0002, r = 0.58, BF10 = 4.68e10, 2-tailed; see Fig. 1 blue section) and reported almost no ability to generate visual sensory details when actively remembering past events (memory vividness on the EMIQ; Mann-Whitney U = 2,186.5, p < 0.0002, r = 0.81, BF10 = 1.01e15, 2-tailed; see Fig. 1 blue section and Fig. S2 in Supplementary Information) compared to participants in control group 1. However, these self-reported reductions in reliving events were not confined to the past, with aphantasics as a group also reporting a near total inability to imagine future hypothetical events in any sensory detail (Future Events component of the SAM; Mann-Whitney U = 7,469.5, p < 0.0002, r = 0.63, BF10 = 2.97e10, 2-tailed; see Fig. 1 blue section and Fig. S2). Self-reported factual (or semantic) memory, which is traditionally thought to provide a kind of ‘scaffold’ for event memories more widely23, also appeared to be lower in individuals unable to visualise compared to controls (Factual Memory component of the SAM; Mann-Whitney U = 18,601.5, p < 0.0002, r = 0.27, BF10 = 156,732.50, 2-tailed; see Fig. 1 blue section and Fig. S2), although this effect was of a lower magnitude than the memory reductions reported above (see Fig. 1 blue section and Table S7 in Supplementary Information). The fourth scale component of the SAM (Spatial Memory) is grouped with the Spatial Imagery component of the OSIQ in results below (see Spatial Ability Results).
Trauma response results
Our data did not directly support the hypothesis that visual imagery absence might protect aphantasic individuals from trauma symptomology in response to stressful life events, with the aphantasia group scoring comparatively to control group 1 on the PCL-5 overall (total PCL-5 scores; Mann-Whitney U = 27,515, p = 0.776, r = 0.01, BF10 = 0.12, 2-tailed; see Fig. 1 grey section and Figure S3 in Supplementary Information). An analysis of group differences on the four sub-components of this scale (Intrusions, Cognition and Mood, Avoidance, and Arousal) also revealed that there were no significant differences between the groups in reports of emotional arousal and reactivity associated with remembering stressful past events (Mann-Whitney U = 27,240, p = 0.924, r = 0.00, BF10 = 0.11, 2-tailed; see Fig. 2b and Fig. S3). Compared to participants with visual imagery, individuals with aphantasia appeared to report fewer recurrent and involuntary memory intrusions (Mann-Whitney U = 22,739, p = 0.002, r = 0.14, BF10 = 14.85, 2-tailed; see Fig. 2b and Fig. S3), lower engagement in avoidance behaviours (Mann-Whitney U = 23,164.5, p = 0.006, r = 0.13, BF10 = 2.13, 2-tailed; see Fig. 2b and Fig. S3), and greater negative changes in cognition and mood (Mann-Whitney U = 30,960, p = 0.008, r = 0.12, BF10 = 12.99, 2-tailed; see Fig. 2b and Fig. S3) in response to stressful life events, although none of these group differences survived Bonferroni correction for multiple comparisons, and effect size estimates were small (r = 0.12–.14; see Table S7 in Supplementary Information). Interestingly, however, Bayesian analyses indicated strong evidence in favour of group differences on the Intrusions (BF10 = 14.85) and Cognition and Mood (BF10 = 12.99) sub-scales of the PCL-5 reported above.
Day and night dream results
Here we found that although there was little evidence for or against (BF10 = 1.93 and BF01 = 0.518) a difference between groups in the reported frequency of day-dreaming (Mann-Whitney U = 23,001.5, p = 0.005, r = 0.13, 2-tailed, non-significant after Bonferroni correction; see Fig. 1 teal section and Figure S4 in Supplementary Information), aphantasic individuals did report experiencing significantly fewer night dreams than controls (Imaginal Process Inventory (IPI); Mann-Whitney U = 15,828.5, p < 0.0002, r = 0.37, BF10 = 4.24e6, 2-tailed; see Fig. 1 teal section and Fig. S4). Interestingly, the reported qualitative content of these night dreams also differed between groups as measured by the SERS. Dream reports for aphantasic individuals reinforce a model of aphantasia as being primarily characterised by sensory deficits (Sensory; Mann-Whitney U = 15,087.5, p < 0.0002, 0.38, BF10 = 5.46e6, 2-tailed) across all dream modalities (including olfactory, tactile, taste and auditory domains; see Fig. 2c and Fig. S4). Interestingly, aphantasic individuals also reported experiencing lower awareness and control during their dreams (Lucidity; Mann Whitney U = 19,473.0, p < 0.0002, r = 0.25, BF10 = 1902.01, 2-tailed). We found some evidence that the dreams aphantasic participants report are characterised by less vivid emotions (Affective; Mann Whitney U = 23,463.0, p = 0.013, non-significant after Bonferroni correction, r = 0.11, BF10 = 9.01, 2-tailed), and a less clear dreamer perspective (Perspective (PSP); Mann Whitney U = 22,070.5, p = 0.0004, r = 0.16, non-significant after Bonferroni correction, BF10 = 127.28, 2-tailed) compared to participants in control group 1. However, there were no significant differences between the aphantasia group and control group 1 in the experience of within-dream cognition (e.g. planning or remembering (Cognitive); Mann Whitney U = 24,592.0, p = 0.085, r = 0.08, BF10 = 1.05, 2-tailed) or the details of dreams’ spatial features (Spatial Complexity (SC); Mann Whitney U = 24,697.0, p = 0.092, r = 0.08, BF10 = 0.31, 2-tailed). Interestingly, the only question on the SERS for which aphantasics scored significantly higher than control group 1 participants was an item in the Cognitive domain (see Fig. 2c) which asks how much time participants spent thinking during their dreams (Mann-Whitney U = 34,401.5, p < 0.0002, BF10 = 3.53e3), which accords well with a reduction in the sensory qualities of dreams in aphantasia in favour of semanticised contents.
Spatial ability results
Aphantasic participants reported slightly lower spatial imagery ability on the spatial sub-component of the OSIQ when compared to control group 1 (Mann-Whitney U = 24,462, p = 0.001, r = 0.15, BF10 = 14.65, 2-tailed; see Fig. 1 purple section and Figure S5 in Supplementary Information), although this effect was not significant after Bonferroni correction. Additionally, the scores of aphantasic individuals on the Spatial Memory component of the SAM (which includes items measuring reported spatial navigation and naturalistic spatial memory ability) were not significantly different from controls (SAM; Mann-Whitney U = 24,720, p = 0.1, r = 0.08, BF10 = 0.23, 2-tailed; see Fig. 1 purple section and Fig. S5). These results demonstrate that overall there were no consistent differences in reported spatial abilities between aphantasic individuals and participants in control group 1.
Control Group 2: Replication Analysis
Although control group 1 was age-matched, it featured a higher ratio of males to females (see Table S1) in contrast to our aphantasic sample (which comprised of more females than males). Some of the variables included in this study (such as spatial ability and PTSD susceptibility) are known to be influenced by gender. To address this potential issue, we ran a replication analysis with a second control group of first-year undergraduate psychology students using the same experimental design (their raw data is depicted alongside our original control group and aphantasic sample in Figures S1–5).
Participants in our second control group (n = 193) were recruited from a sample of undergraduate psychology students at the University of New South Wales, and completed the study in exchange for course credit. All participants in this second control group were included in final analysis (with no exclusions). These participants (mean age = 19.33 years, SD = 3.69, range = 17–55 years) were not matched on mean age with our aphantasic sample (mean age difference = 14.6 years, p < 0.01, BF10 = 1.23e10), but instead featured a higher proportion of females to males (73% females, compared to 48% females in our aphantasic sample and 35% females in control group 1 (our main control group of MTurk responders).
Comparison with this second control group revealed a similar overall pattern of group differences to those reported above, with few effect changes in imagery and memory related domains in particular (see Figures S1–5 and Tables S2–6 in Supplementary Information for a comparison of test results, as well as Table S7 for a comparison of effect sizes). Aphantasic participants scored significantly lower than control group 2 on all outcomes of the imagery and episodic memory questionnaires (all p < 0.0002, all r > 0.52, all BF10 > 1.42e8) with the exception of the factual memory component of the SAM (which was no longer significantly lower in aphantasics when compared to control group 2 after controlling for multiple comparisons; Mann-Whitney U = 21,496.0, p = 0.002, r = 0.14, BF10 = 3.196, 2-tailed).
Although our Bayes analysis suggested strong evidence for higher total PCL-5 scores in control group 2 compared to the aphantasic group (Mann-Whitney U = 21,464.0, p = 0.002, r = 0.14, BF10 = 12.76, 2-tailed), this effect was not significant after Bonferroni correction. However, the previously non-significant reduction in memory intrusions amongst aphantasic participants (compared to control group 1) was much stronger in this second group comparison (Mann-Whitney U = 15,134.5, p < 0.0002, r = 0.35, BF10 = 2.20e7, 2-tailed), as were lower reports of avoidance behaviours by aphantasic individuals compared to control group 2 (Mann-Whitney U = 18,494.5, p < 0.0002, r = 0.24, BF10 = 2494.67, 2-tailed). Compared to control group 2, however, aphantasic participants did not report significantly higher negative cognition and mood (Mann-Whitney U = 25,827.5, p = 0.97, r = 0.00, BF10 = 0.12, 2-tailed) or arousal (Mann-Whitney U = 25,517.0, p = 0.12, r = 0.07, BF10 = 0.34, 2-tailed) in response to stressful life events, in line with our main control group 1 comparisons.
Individuals with aphantasia reported significantly fewer night dreams than control group 2 (Mann-Whitney U = 17,156.0, p < 0.0002, r = 0.74, BF10 = 21,124.12, 2-tailed). However, they also reported significantly less frequent mind-wandering compared to participants in control group 2 (Mann-Whitney U = 19,271.5, p < 0.0002, r = 0.29, BF10 = 397.04, 2-tailed), in contrast to the results of our main analysis (which revealed no significant differences in mind-wandering reports between the aphantasic group and control group 1). Also in contrast to our initial dreaming results, aphantasic participants scored significantly lower than control group 2 on all components of the SERS (Sensory, Affective, Cognitive, Spatial Complexity, Perspective and Lucidity; all p < 0.0002, all r > 0.71, all BF10 > 1.56e7), including on some domains where there were no significant differences between aphantasic participants and age-matched participants in control group 1 (see Fig. S4 and Table S5). However, these findings may be partially explained by age-related decline in dream frequency and subjective recall24.
Lastly, there were no significant differences in reported spatial imagery ability on the OSIQ (Mann-Whitney U = 22,635.5, p = 0.03, r = 0.10, BF10 = 0.88, 2-tailed) or spatial navigation ability on the SAM (Mann-Whitney U = 23,760.5, p = 0.15, r = 0.07, BF10 = 0.23, 2-tailed) between the aphantasic group and control group 2, reinforcing our initial results as well as previous findings of preserved spatial (but not object) imagery in aphantasic participant samples10,22.