Memory is inherently entangled with prediction and planning. Flexible
behavior in biological and artificial agents depends on the interplay of
learning from the past and predicting the future in ever-changing environments.
This chapter reviews computational, behavioral, and neural evidence suggesting
these processes rely on learning the relational structure of experiences, known
as cognitive maps, and draws two key takeaways. First, that these memory
structures are organized as multiscale, compact predictive representations in
hippocampal and prefrontal cortex, or PFC, hierarchies. Second, we argue that
such predictive memory structures are crucial to the complementary functions of
the hippocampus and PFC, both for enabling a recall of detailed and coherent
past episodes as well as generalizing experiences at varying scales for
efficient prediction and planning. These insights advance our understanding of
memory and planning mechanisms in the brain and hold significant implications
for advancing artificial intelligence systems.

In this chapter, we delve into the relationship between memory, prediction, and planning. We explore the interdisciplinary nature of these concepts and analyze evidence from computational models, behavioral studies, and neural research to gain a comprehensive understanding of how these processes interact.

The Interplay of Memory, Prediction, and Planning

Memory is not simply a passive process of storing past experiences. Instead, it is intimately entwined with prediction and planning. Flexible behavior in both biological and artificial agents hinges on the ability to learn from past events and use this knowledge to make predictions about the future. This adaptive behavior is essential for navigating ever-changing environments.

Learning the Relational Structure: Cognitive Maps

One key insight we derive from the evidence is that memory structures rely on learning the relational structure of experiences. These structures, often referred to as cognitive maps, provide a framework for organizing and representing past events. These cognitive maps allow for efficient recall of past episodes as well as generalization of experiences at different scales, enabling effective prediction and planning.

Multiscale Predictive Representations in the Brain

The evidence suggests that these memory structures are organized as multiscale, compact predictive representations in two key brain regions: the hippocampus and prefrontal cortex (PFC). The hippocampus plays a crucial role in forming detailed and coherent representations of specific episodes, while the PFC facilitates generalization and prediction across different contexts.

The Complementary Functions of the Hippocampus and PFC

We argue that the interplay between the hippocampus and PFC is essential for robust memory and planning capabilities. The hippocampus enables the recall of rich episodic memories, providing specific details about past experiences. On the other hand, the PFC aids in generalizing these memories to make predictions and plans that are applicable to a wide range of situations.

Implications for Artificial Intelligence Systems

Understanding the mechanisms underlying memory and planning in the brain also has significant implications for artificial intelligence systems. By incorporating these insights, we can design AI systems that possess more human-like memory and planning capabilities. This can enhance their ability to navigate complex and dynamic environments, leading to more intelligent and adaptive AI systems.


The study of memory, prediction, and planning is a multi-disciplinary endeavor that brings together computational, behavioral, and neural perspectives. By unraveling the intricacies of these processes and their underlying neural mechanisms, we gain valuable insights that advance our understanding of both the human brain and artificial intelligence. These insights have far-reaching implications for developing more intelligent and adaptive systems in various domains.

Read the original article