Event networks and the neural representations that support real-world memory

点击翻译按钮获取中文摘要

Project Summary Memory systems evolved to inform the continual learning and decision making of organisms as they explore and engage with an enormously complicated world. Humans in particular have a remarkable ability to recount complex sequences of events: we can easily reconstruct a narrative about the past hour or day purely from memory. In such real-world remembering, semantic and causal associations become exceedingly important, defining a web of relational connections across time to guide recall. For example, your day might contain two "hub" events: a dinner party that requires visiting several shops to pick up supplies, and a morning phone call saying that your child has a fever and needs to go home; each spawns a multitude of events that make up your day. Rich associations among these moments form an “event network” whose local and global properties shape recall; your decisions guide how each event will unfold. While studies show that relations between simple items are important for memory organization and its accompanying neural computations, no existing models consider the higher-order structure of networks composed from inter-related naturalistic events. Even among naturalistic studies, most use passively-viewed movies or stories; participants have no choices to make or goals to pursue. This lack of attention to the higher-order network properties and volitional aspects of real- world experiences has hindered efforts to identify the cortical dynamics that underlie ecologically meaningful memory processes. We seek to understand how memory encoding and retrieval of realistic events is implemented, in terms of cortical representations and interactions between brain systems. Doing so requires paradigms with two critical attributes. First, the stimuli must be sufficiently complex. Memory researchers have long focused on reductive scenarios with isolated stimuli that intentionally destroy semantic and causal connections. In contrast, our experiments use realistic events that are richly associated with each other and will naturally generate a diversity of event network structures. Second, participants must take an active role in creating their memories. Organisms in the real world can volitionally interact with their input stream: at a crowded party, you can choose to explore the kitchen or the living room, talk to the biologist or the musician, leave early or stay until dawn. We will test how participants' volitional behaviors, as they interact with and actively seek information about their environment, shape event networks and neural representations of events. Altogether, these experiments will provide novel frameworks and tools to examine how emergent higher-order structure in natural experiences governs the neural mechanisms underlying encoding and recall. By advancing the level of ecological validity and stimulus complexity in human memory research, we expect to help uncover brain-behavior relationships not apparent in simpler paradigms, and increase the translatability of laboratory findings to real-world applications.