Quantifying aphantasia through drawing: Those without visual imagery show deficits in object but not spatial memory

Abstract

Congenital aphantasia is a recently characterized variation of experience defined by the inability to form voluntary visual imagery, in individuals who are otherwise high performing. Because of this specific deficit to visual imagery, individuals with aphantasia serve as an ideal group for probing the nature of representations in visual memory, particularly the interplay of object, spatial, and symbolic information. Here, we conducted a large-scale online study of aphantasia and revealed a dissociation in object and spatial content in their memory representations. Sixty-one individuals with aphantasia and matched controls with typical imagery studied real-world scene images, and were asked to draw them from memory, and then later copy them during a matched perceptual condition. Drawings were objectively quantified by 2,795 online scorers for object and spatial details. Aphantasic participants recalled significantly fewer objects than controls, with less color in their drawings, and an increased reliance on verbal scaffolding. However, aphantasic participants showed high spatial accuracy equivalent to controls, and made significantly fewer memory errors. These differences between groups only manifested during recall, with no differences between groups during the matched perceptual condition. This object-specific memory impairment in individuals with aphantasia provides evidence for separate systems in memory that support object versus spatial information. The study also provides an important experimental validation for the existence of aphantasia as a variation in human imagery experience.

Keywords: False memory; Memory recall; Mental imagery; Object information; Spatial information.

Fig. 1.

The experimental design of the online experiment. Participants 1) studied three separate scene photographs presented sequentially, 2) drew them from memory, 3) completed a recognition task, 4) copied the images while viewing them, and then 5) filled out the VVIQ and OSIQ questionnaires in addition to demographics questions. The whole experiment took approximately 30 minutes.

Figure 2.. Experimental paradigm and basic demographics.

a) b) (Left) A histogram of the distribution of participants across the VVIQ. Aphantasic participants were selected as those scoring 25 and below (N=61) and controls were selected as those scoring 40 and above (N=52), while those in between were removed from the analyses (N=8). While the range of the VVIQ is from 16 to 80, some participants (N=10 out of 121 total) skipped 1–3 questions, leading to some participants scoring below 16. These skipped questions did not affect group membership. (Middle) A scatterplot of total VVIQ score plotted against total OSIQ Object component score for participants meeting criterion. Each point represents a participant, with aphantasic participants in blue and controls in red. There was a significant difference in OSIQ Object score between the two groups. (Right) A scatterplot of total VVIQ score plotted against OSIQ Spatial component score. There was no difference in OSIQ Spatial score between the two groups. Both the OSIQ Object component and Spatial components have a range of 15 to 75 points.

Figure 3.. Example drawings.

Example drawings made by aphantasic and control participants from memory and perception (i.e., copying the image) showing the range of performance. The memory and perception drawings connected by arrows are from the same participant, and every row is from a different participant. Low memory examples show participants who drew the fewest from memory but the most from perception. High memory examples show participants who drew the highest amounts of detail from both memory and perception. These examples are all circled in the scatterplot of Fig. 4. The key question is whether there are meaningful differences between these two sets of participants’ drawings.

Figure 4.. Comparison of object information in drawings between aphantasic and control participants.

(Left) A scatterplot of each participant as a point, showing average number of objects drawn from memory across the three images (x-axis), versus average number of objects drawn from perception across the three images (y-axis). Aphantasic participants are in blue, while control participants are in red. The bright blue circle indicates average aphantasic performance, while the bright red circle indicates average control performance, with crosshairs for both indicating standard error of the mean for memory and perception respectively. Histograms on the axes show the number of participants who drew each number of objects. Controls drew significantly more objects from memory, although with a tendency towards fewer from perception. The circled light blue and red points are the participants with the lowest memory performance shown in Fig. 3, while the circled dark blue and red points are the participants with the highest memory performance shown in Fig. 3. (Right) Heatmaps of which objects for each image tended to be drawn more by controls (red) or aphantasic participants (blue). Pixel value represents the proportion of control participants who drew that object in the image subtracted by the proportion of aphantasic participants who drew that object (with a range of −1 to 1). Controls remembered more objects (i.e., there is more red in the memory heatmaps), even though aphantasic participants tended to copy more objects (i.e., there is more blue in the perception heatmaps).

Figure 5.. Average object locations and sizes recalled by aphantasics and controls.

Average object locations and sizes for memory drawings of four of the main objects from each image, made by aphantasic participants (solid lines) and control participants (dashed lines). Even though these objects were drawn from memory, their location and size accuracy was still very high. Importantly, aphantasic and control participants showed no significant differences in object location or size accuracy.

Figure 6.. False object memories in the drawings.

Examples of the false object memories made by participants in their memory drawings, with the inaccurate objects circled. Control participants made significantly more errors, with only 3 out of 176 total aphantasic drawings containing a falsely remembered object. Note that all aphantasic errors were also transpositions from other images.