Effective connectivity of the anterior hippocampus predicts recollection confidence during natural memory retrieval

Abstract

Human interactions with the world are influenced by memories of recent events. This effect, often triggered by perceptual cues, occurs naturally and without conscious effort. However, the neuroscience of involuntary memory in a dynamic milieu has received much less attention than the mechanisms of voluntary retrieval with deliberate purpose. Here, we investigate the neural processes driven by naturalistic cues that relate to, and presumably trigger the retrieval of recent experiences. Viewing the continuation of recently viewed clips evokes greater bilateral activation in anterior hippocampus, precuneus and angular gyrus than naïve clips. While these regions manifest reciprocal connectivity, continued viewing specifically modulates the effective connectivity from the anterior hippocampus to the precuneus. The strength of this modulation predicts participants' confidence in later voluntary recall of news details. Our study reveals network mechanisms of dynamic, involuntary memory retrieval and its relevance to metacognition in a rich context resembling everyday life.

Fig. 1

GLM results of news clip viewing. a Brain regions with statistically significant activation during first viewing, b continued-viewing, c naïve viewing. d Brain regions with greater activation to the continuing viewing than naïve viewing. e Average T statistics in left hippocampus (1–4 represents four hippocampal parcels from posterior to anterior hippocampus; see Methods, Supplementary Figure 14) and DMN (P: precuneus, L_A: left angular gyrus, R_A: right angular gyrus, mPFC: medial prefrontal cortex) across participants. Significant voxels were identified using a FWE cluster-corrected threshold p < 0.001, with a voxel height defined with p < 0.001. Error bars signify SEM

Fig. 2

DCM analyses of endogenous connectivity for second viewing. a Specification of the model space in terms of four sets of connectivity families, including connections between visual and precuneus, angular gyrus and precuneus, precuneus and hippocampus, angular gyrus and hippocampus, resulting in 72 competing models. For all the models, the visual ROI receives the driving input. b Exceedance probability for the DCM model families in the BMS, including visual–precuneus families, angular gyrus–precuneus families, precuneus–hippocampus families, and angular gyrus–hippocampus families. The label of horizontal axis for each box corresponds to the label in (a). Bidirectional families are given first in each box. c Results of the BMS for all the 72 models. HIPP = Hippocampus; Precun = Precuneus; Angular = Angular gyrus

Fig. 3

DCM analyses of modulations for the second news viewing. The models are specified by adding modulations corresponding to the continued-viewing effect. a Identifications of the model space in terms of three sets of modulator families, including modulator acting on connections between angular gyrus and precuneus, angular gyrus and hippocampus, precuneus and hippocampus, resulting in 64 competing models. b Exceedance probability (right) for the DCM model families in the BMS, including angular gyrus–precuneus families, and angular gyrus–hippocampus families, precuneus–hippocampus families. The label of horizontal axis for each box corresponds to the label in (a). c Results of the BMS across all the 64 models. HIPP = Hippocampus; Precun = Precuneus; Angular = Angular gyrus

Fig. 4

Winning model of DCM analyses considering modulations effect. This is the winning model of the space defined by adding modulators to the intrinsic connectivity model. Connectivity, driving input and modulator are coded in the thickness of line, and the size of arrow and circle, respectively, representing their effect size (c: continuing news clip, n: naïve news clip). HIPP = Hippocampus; Precun = Precuneus; Angular = Angular gyrus

Fig. 5

Correlation between DCM parameter and behaviour data. Correlation between the DCM parameter of the modulation of the effective connectivity from the hippocampus to the precuneus and (a) the number of confident answers, (b) the number of accurate answers derived from news recall task