Flexible network community organization during the encoding and retrieval of spatiotemporal episodic memories

Abstract

Memory encoding and retrieval involve distinct interactions between multiple brain areas, yet the flexible structure of corresponding large-scale networks during such memory processing remains unclear. Using functional magnetic resonance imaging, we employed a spatiotemporal encoding and retrieval task, detecting functional community structure across the multiple components of our task. Consistent with past work, we identified a set of stable subnetworks, mostly belonging to primary motor and sensory cortices but also identified a subset of flexible hubs, mostly belonging to higher association areas. These “mover” hubs changed connectivity patterns across spatial and temporal memory encoding and retrieval, engaging in an integrative role within the network. Global encoding network and subnetwork dissimilarity predicted retrieval performance. Together, our findings emphasize the importance of flexible network allegiance among some hubs and the importance of network reconfiguration to human episodic memory.

Prefrontal cortical activity predicts the extra-place field spiking of hippocampal place cells

“…Our results also suggest that information exchange between cortex and hippocampus may occur frequently during active behavior. This extends previous findings of hippocampal-prefrontal coupling from imaging (Squire, Ojemann et al 1992, Buckner, Petersen et al 1995, Schacter, Alpert et al 1996, Polyn, Natu et al 2005, St Jacques, Kragel et al 2011, Rugg and Vilberg 2013, Schedlbauer and Ekstrom 2019) and neural recording experiments (Tomita, Ohbayashi et al 1999, Kyd and Bilkey 2003, Hyman, Zilli et al 2005, Jones and Wilson 2005, Benchenane, Peyrache et al 2010, Sigurdsson, Stark et al 2010, Hok, Chah et al 2013, Place, Farovik et al 2016, Guise and Shapiro 2017, Myroshnychenko, Seamans et al 2017, Zielinski, Shin et al 2019) on the role of prefrontal cortex in modulating both cortical and subcortical structures during mnemonic processes (Tomita, Ohbayashi et al 1999, Simons and Spiers 2003, St Jacques, Kragel et al 2011, Eichenbaum 2017). Our results also complement findings demonstrating coherent spiking activity patterns across hippocampus and PFC in the context of both SWRs and locomotion-associated spiking (Jadhav, Rothschild et al 2016, Shin, Tang et al 2019, Zielinski, Shin et al 2019).…”

“…We were therefore surprised to find PFC spiking can predict whether isolated spiking will occur up to 4-8 theta cycles, or approximately 500ms -1s, later, an interval much longer than what is needed for direct information transfer. Although the channel for communication between PFC and hippocampus may have a short latency, our results suggest the expression of isolated hippocampal spiking likely involves coordinated activity between these regions (Squire, Ojemann et al 1992, Buckner, Petersen et al 1995, Schacter, Alpert et al 1996 that evolves over time (Polyn, Natu et al 2005, Rugg and Vilberg 2013, Schedlbauer and Ekstrom 2019. This is consistent with human imaging studies that show cortical activity change can precede memory recall on the order of seconds (Polyn, Natu et al 2005).…”

Amygdala and ventral tegmental area differentially interact with hippocampus and cortical medial temporal lobe during rest in humans

“…In the ensuing logic, VTA modulation over the hippocampus may emerge only in affective states that engage VTA, which has been previously been shown in the context of reward motivation (Adcock et al, 2006; Gruber, Watrous, Ekstrom, Ranganath, & Otten, 2013; Murty & Adcock, 2014; Wolosin, Zeithamova, & Preston, 2013), novelty (Wittmann, Daw, Seymour, & Dolan, 2008), and curiosity (Gruber, Gelman, & Ranganath, 2014). Further from a mnemonic perspective, this research compliments recent research showing that regions within motor and sensory cortex dynamically shift as a function of mnemonic state to better support long‐term memory (Schedlbauer & Ekstrom, 2019).…”

Reconfiguration dynamics of a language-and-memory network in healthy participants and patients with temporal lobe epilepsy

“…These modules are sparsely connected with other communities via inter-module connections that provide the integration property ( Fornito et al, 2016 , Guimerà and Amaral, 2005 , Rubinov and Sporns, 2010 ). Nodes that are highly interconnected within their communities, but not so strongly to other communities are called provincial hubs and they support segregation ( Bertolero et al, 2015 , Meunier et al, 2009 , Schedlbauer and Ekstrom, 2019 ). The integration between modules is based on the connector nodes that are highly connected with other communities and can be divided into satellites and connector hubs depending on their status within their own community ( Bertolero et al, 2015 , Bullmore and Sporns, 2012 , Fornito et al, 2016 , Guimerà and Amaral, 2005 , Meunier et al, 2010 ).…”

“…The integration between modules is based on the connector nodes that are highly connected with other communities and can be divided into satellites and connector hubs depending on their status within their own community ( Bertolero et al, 2015 , Bullmore and Sporns, 2012 , Fornito et al, 2016 , Guimerà and Amaral, 2005 , Meunier et al, 2010 ). The difference is that connector hubs are, unlike satellites, also highly interconnected within their communities ( Bertolero et al, 2015 , Meunier et al, 2009 , Schedlbauer and Ekstrom, 2019 ).…”