The experience of vivid autobiographical reminiscence is supported by subjective content representations in the precuneus

Abstract

The human posteromedial cortex, which includes core regions of the default mode network (DMN), is thought to play an important role in episodic memory. However, the nature and functional role of representations in these brain regions remain unspecified. Nine participants (all female) wore smartphone devices to record episodes from their daily lives for multiple weeks, each night indicating the personally-salient attributes of each episode. Participants then relived their experiences in an fMRI scanner cued by images from their own lives. Representational Similarity Analysis revealed a broad network, including parts of the DMN, that represented personal semantics during autobiographical reminiscence. Within this network, activity in the right precuneus reflected more detailed representations of subjective contents during vivid relative to non-vivid, recollection. Our results suggest a more specific mechanism underlying the phenomenology of vivid autobiographical reminiscence, supported by rich subjective content representations in the precuneus, a hub of the DMN previously implicated in metacognitive evaluations during memory retrieval.

Figure 1

(a) The phone is worn around the neck with its camera exposed as shown. (b) A word cloud of the tags associated with the stimuli used in the fMRI experiment across all participants. Relative font sizes indicate relative frequencies of the tags while color and orientation are merely for visualization purposes.

Figure 2

Common neighbor ratio comparison of image representations. The color correlogram representation achieves the best congruence between visual and temporal proximity of nearest neighbors (k).

Figure 3

Depiction of the fMRI experiment and Representational Similarity Analysis (RSA). Participants are shown images from their own lives and are instructed to relive the associated experiences. The neural activity during this reminiscence period is analyzed using RSA to investigate whether distances between neural patterns (NeuralD_ij) corresponding to pairs of image cues (an example of such a pair is shown) relate to distances between the corresponding sets of semantic tags (HammD_ij). After the reminiscence period, participants indicate whether they remember the event and then report the vividness of their recollective experience.

Figure 4

(a) Normalized pointwise mutual information (NPMI) between all pairs of tags, computed across participants. (b) A network of tags with NPMI > 0.2.

Figure 5

(a) The network of regions involved in the representation of general personal semantics as identified by the RSA analysis in Eq. 5 (corresponding to the Hamm term). Four different views (left, back, right, top) of a glass brain are shown. A full 3D view of the network can be seen in Supplementary Movie S1. See Table 2 for a complete list of regions with at least 10 voxels in the general personal semantic network. (b) The network of regions involved in the representation of personal semantics during vivid reminiscence as identified by the RSA analysis in Eq. 6 (corresponding to the Hamm term). See Table 3 for a complete list of regions with at least 10 voxels in the vivid personal semantics network. The same views presented in (a) are shown and comparing the two networks reveals that the vivid reminiscence network is a subset of the more general personal semantics network identified in (a). A full 3D view of the network can be seen in Supplementary Movie S2.

Figure 6

The right precuneus represents personal semantics during vivid reminiscence but to a lesser extent during non-vivid reminiscence (Eq. 6, corresponding to the conjunction between the Hamm and Hamm * Vivid terms, see Methods).

Figure 7

The slopes of the regression lines in Eq. 6 describing the relationship between neural distances and Hamming distances between the tag sets in a sphere of radius 7.5 mm around the peak voxel in the right precuneus. (a) The colored lines show individual participants’ regression lines for the relationship between Hamming distance and neural distance for vividly remembered pairs of images after accounting for the contribution from other independent variables in Eq. 6 (i.e., the partial residual). The slope of the solid black line is the mean over the individual regression lines. (b) The relationship between Hamming distance and neural distance for the less vividly remembered pairs of images after accounting for the contribution from other independent variables in Eq. 6. Individual participants’ plots with partial residuals overlaid are presented in Fig. 8.

Figure 8

Individual participant partial residual plots of the Neural distance ~ Hamming distance relationship for vivid (left panel) and non-vivid (right panel) pairs.