Temporal integration of narrative information in a hippocampal amnesic patient

Abstract

Default network regions appear to integrate information over time windows of 30 seconds or more during narrative listening. Does this long-timescale capability require the hippocampus? Amnesic behavior suggests that the hippocampus may not be needed for online processing when input is continuous and semantically rich: amnesics can participate in conversations and tell stories spanning minutes, and when tested immediately on recently heard prose their performance is relatively preserved. We hypothesized that default network regions can integrate the semantically coherent information of a narrative across long time windows, even in the absence of the hippocampus. To test this prediction, we measured BOLD activity in the brain of a hippocampal amnesic patient (D. A.) and healthy control participants while they listened to a seven-minute narrative. The narrative was played either in its intact form, or as a paragraphscrambled version, which has been previously shown to interfere with the long-range temporal dependencies in default network activity. In the intact story condition, D. A.'s moment-by-moment BOLD activity spatial patterns were similar to those of controls in low-level auditory cortex as well as in some high-level default network regions (including lateral and medial posterior parietal cortex). Moreover, as in controls, D. A.'s response patterns in medial and lateral posterior parietal cortex were disrupted when paragraphs of the story were presented in a shuffled order, suggesting that activity in these areas did depend on information from 30 seconds or more in the past. Together, these results suggest that some default network cortical areas can integrate information across long timescales, even in the absence of the hippocampus.People with hippocampal damage are profoundly impaired in recalling information after a distraction (Milner, 1966), or "as soon as their attention shifts to a new topic" (Milner, 2005). At the same time, these amnesic individuals are able to engage in conversation, retain near-normal immediate recall for prose passages (Baddeley and Wilson, 2002), and tell globally coherent stories (Keven et al., 2018;Kurczek and Duff, 2011;Rosenbaum et al., 2009). How is this possible? One potential explanation is that, in the absence of major topic changes or surprises that create a distraction, amnesic individuals are able to rely on default network cortical regions for retention of semantically-rich information across time. Default network areas are proposed to carry slowly-changing information during continuous natural input such as stories and conversation (Hasson et al., 2015). In healthy people, these areas are functionally coupled to the hippocampus and may work together to accumulate, maintain, and integrate information across events. However, it is an open question whether the long-timescale capability of default network regions depends critically on interactions with the hippocampus .In this study, we investigated whether default network areas can integrate information over tens of se...

When to retrieve and encode episodic memories: a neural network model of hippocampal-cortical interaction

“…We find that, if the model is allowed to continuously maintain information in working memory (i.e., there are no distractions/interruptions that clear out the working memory state), the model is sensitive to stimulus history, even if the hippocampus is completely inactivated. This result fits with the results of a recent patient study (Zuo et al, 2020) that showed that an amnesic patient’s brain response was sensitive to “scrambling” which parts of a movie preceded the current part, as well as other, more anecdotal results showing that hippocampal patients can comprehend narratives so long as they are not interrupted (Milner et al, 1968).…”

“…Inter-subject analyses are widely used in naturalistic studies, such as the studies by J. Chen et al (2016) and Zuo et al (2020) that we discuss in Simulations 1, 6, and 7 . Comparing data across subjects requires some form of alignment.…”

“…The specific goal of this simulation was to model the findings of a recent study by Zuo et al (2020), which found that a patient with a hippocampal lesion had significant (above zero) scrambling-sensitivity in the DMN. Specifically, Zuo et al (2020) computed spatial ISC between controls and a hippocampal patient while they were listening to an audiobook. Spatial ISC at a given moment is the linear correlation between the multivoxel patterns of controls and patients.…”

Mapping the organization and dynamics of the posterior medial network during movie watching

“…In support, prior work suggests that temporal integration of narrative information during movie watching in Dorsal PMN regions may not depend on interactions with the hippocampus (Chen et al, 2016;Zuo et al, 2020). The hippocampus has been previously characterized as a gateway between the PMN and an anterior temporal (AT) network that processes item and emotional information (Ranganath & Ritchey, 2012;Ritchey, Libby, et al, 2015).…”

Narratives: fMRI data for evaluating models of naturalistic language comprehension

“…The following subjects received the “Pie Man” stimulus on two separate occasions (specified by or in the BIDS file naming convention): , , , , , , , , , , , , , . We recommend excluding subjects (both and ), (), (), , , , , , and (as specified in the file in the directory) based on criteria explained in the “Intersubject correlation” section of “Data validation.” Subsets of the “Pie Man” data have been previously reported in numerous publications (Lerner et al, 2011, 2014; Ben-Yakov et al, 2012; Regev et al, 2013; Stephens et al, 2013; Chen et al, 2015; Simony et al, 2016; Baldassano et al, 2017; Liu et al, 2017; Vodrahalli et al, 2017; Yeshurun et al, 2017a; Zuo et al, 2020), and additional datasets not shared here have been collected using the “Pie Man” auditory story stimulus (e.g. Gross et al, 2013; Blank & Fedorenko, 2017; Iotzov et al, 2017).…”