NIMH/NIH Silvio O. Conte Center for Neuroscience Research

Prefrontal and Medial-Temporal Interactions in Memory


take 2


A fundamental challenge in neuroscience is understanding of how major brain areas operate together as a system to support high level cognitive functions.  In particular, one system of great importance to the understanding of mental disorders involves the prefrontal cortex (PFC) and the medial temporal lobe (MTL) that each contribute to high level functions in memory and cognition.  In recent years, there has been significant progress in revealing the individual functions of areas within the PFC and MTL, and some successes in showing that these areas work together.  However, there is a paucity of knowledge about the mechanisms by which PFC and MTL interact in the service of memory and cognition.  Here we bring together a group of investigators who have led research on PFC and MTL, or both, using diverse approaches and different species.  We have collaborated to develop an hypothesis of PFC-MTL interactions and aim to integrate our strengths towards revealing the mechanisms of that interaction, and in so doing, pioneer a true systems level understanding of memory and cognition.  Our strategy combines theories of PFC and MTL function generated by our group with coordinated experimental analyses that converge on a common behavioral paradigm for exploring encoding and retrieval operations in context-guided associative memory.  Experimental projects will examine how PFC and MTL areas contribute individually and interactively, using diverse and intersecting approaches: in humans, using both neuropsychological studies of brain damaged individuals and fMRI on normal human subjects, in monkeys using neurophysiological studies of single neuron activity and local field potentials, in rats using combined neurophysiological and reversible inactivation studies of non-spatial and spatial memory. The findings will be integrated in two key ways, by direct comparison of different data modalities through  analysis  and by convergence into a computational model that provides a comprehensive and unified account of PFC-MTL interactions.  The Center will also pursue several directly related educational, dissemination, and outreach goals.

Figure Legend: Anatomy of PFC and MTL areas involved in memory, indicating some of their prominent connections.

Project 1

Neuropsychological investigations of PFC-MTL interactions in humans.

Neal J. Cohen

University of Illinois at Urban-Champaign

Beckman Institute


Joel Voss
Northwestern University Feinberg School of Medicine

Overview:  This project combines neuropsychological, neuroimaging, and eye tracking approaches in order to study the functional interactions of PFC and the hippocampus in supporting richly conditional behavior in humans. The experiments test our hypothesis that the hippocampus is critically involved in relational memory representations whereas PFC is involved in more abstract context-guided associative rules. Neuropsychological studies will provide evidence about the necessity of PFC and MTL regions in relational memory and context-dependent associations, and neuroimaging studies will provide evidence about the nature and timing of functional interactions between these regions. For each study, performance assessments will include not only explicit behavioral judgments but also eye movement-based assessment of memory, pioneered in our laboratory. This approach affords sensitive, implicit measures of the strength of relational and context-dependent representations, based on preferential viewing patterns, as they change dynamically during each trial and across learning and retention. Experiments start with a “base” task in common with all the other empirical projects of the Center, and then graduate to more elaborate variants that systematically manipulate the amount and complexity of relational information or the complexity and abstractness of the context-dependent associative rules to be learned, in order to better determine the dependency of each of these aspects of memory on PFC, hippocampus, and their functional interactions, and further, on the directionality of their interactions.

Expansion of Scope-Project 1

The original Specific Aims of Project 1 remain unchanged, and are as follows:

Specific Aim 1 To demonstrate the necessary and distinct roles of PFC and MTL, by assessing explicit choice behavior and preferential viewing patterns in patients with selective lesions to VLPFC, DLPFC, OFC, or hippocampus.

Specific Aim 2 To examine PFC-hippocampal interactions in combined fMRI and eye movement studies in healthy individuals along with combined neuropsychological and eye movement studies in patients with selective lesions.

By expanding the scope to include TMS neuromodulatory techniques, we add a new Specific Aim for Year 2:

Specific Aim 3 To examine the necessary role of PFC-MTL interactions in context-dependent association learning using combined fMRI and disruptive TMS studies in healthy individuals to determine the regions that make critical contributions to specific stages of learning and the timecourse of PFC-MTL interactions.

Exploratory Aim: To determine parameters of TMS enhancement stimulation regimens needed to enhance PFC-MTL networks and improve context-dependent association learning.

The new Specific Aim and Exploratory Aim are logical extensions of the original Aims 1 and 2 that takes advantage of newly available technologies to add greater anatomical and temporal precision to our lesion-deficit studies (an expansion of Aim 1) as well as causal inferences to be made from our fMRI studies (an expansion of Aim 2).

Project 2

fMRI investigations of PFC-MTL interactions in humans.

Chantal Stern

Boston University


Overview: This project will combine functional magnetic resonance imaging (fMRI) with functional and effective connectivity methods to examine the interactions between regions of the prefrontal cortex (PFC) and medial temporal lobe (MTL) in humans.  The fMRI paradigms are closely related to the context-based associative memory and context-based inference tasks that will be used in patients with selective PFC or MTL lesions (Project 1), and monkeys (Project 3), and are related to the context-dependent associative memory tasks used in rodents (Projects 4 and 5). The specific aims of Project 2 are:

Specific Aim 1  To examine PFC-MTL interactions in the human during the learning and retrieval of context-based associations.We will use fMRI in combination with functional and effective connectivity methods to examine interactions between the MTL and PFC regions during the learning and retrieval phases of a context-based associative memory task.

Specific Aim 2 To examine PFC-MTL interactions in the human during a context-based inference task. We will use fMRI in combination with functional and effective connectivity methods to examine interactions between the MTL and PFC regions during a context-based inference task.

Specific Aim 3 To examine PFC-MTL interactions in the human during a relational-load task. We will use fMRI in combination with functional and effective connectivity methods to examine interactions between the MTL and PFC regions during a relational-load task, based on the task described in Project 1. The primary comparison will be between a high relational load condition and one with a low relational load.

Project 3

Circuit analysis of PFC and MTL interactions in monkeys

Earl K. Miller

Massachusetts Institute of Technology


Overview. This project is a link between human, monkey, and rodent projects.  It provides insight at the level of the brain’s basic elements (neurons) and the networks that underlie the function of putatively homologous areas in humans (Projects 1 & 2).  At the same time, our neurophysiological data is most comparable to that from the rodent projects (Projects 4 and 5).  Thus, this project is a bridge to the development of animal models of human memory disorders. The specific aims of Project 3 are:

Specific Aim 1 To determine if the PFC provides context information to the MTL during learning and retrieval of position-dependent memories. We will test this for the first time in monkeys.

Specific Aim 2. To determine if the PFC makes memory-based inferences by retrieving memories from the MTL.

Project 4

Circuit analysis of PFC-MTL interactions in rats

Howard Eichenbaum

Boston University


Overview. This project extends our exploration of PFC-MTL interactions to rats and complements the approaches taken in studies on humans and monkeys on context guided associative memory by examining the causal interactions between areas in the PFC-MTL circuit. The Specific Aims of this project are:

Specific Aim 1 We will explore PFC-MTL interactions in rodents by comparing and correlating neural activation patterns in context-guided associative memory.  These experiments will shed light on whether PFC and MTL areas and their interactions are functionally homologous to those in humans and monkeys.

Specific Aim 2 We will determine whether areas in PFC, MTL, and their interactions are critical to context-guided associations by assessing the effects of reversible inactivation of specific subareas, and by crossed PFC-MTL inactivation which tests whether ipsilateral interactions between these areas are critical.

Specific Aim 3 We will employ combinations of recording and reversible inactivation to examine how PFC areas influence neural firing patterns in MTL areas, and vice versa.  These experiments will inform us about causal relations between activity in one area and information processing in another.  The results are essential to modeling the flow of information between components of PFC-MTL circuitry.

Project 5

PFC-MTL functional interactions in spatial memory

Matthew L. Shapiro

Mount Sinai School of Medicine


Overview. The MTL is crucial for memory in humans, monkeys and rats (Eichenbaum & Cohen, 2001), yet a vast literature emphasizes its role in spatial mapping, navigation, and path integration (e.g. McNaughton et al., 2006). We propose that the PFC and MTL interact bidirectionally to support context guided learning and memory retrieval, and that spatial mapping, navigation, and path integration exemplify the outcome of these more general computations. By investigating the same PFC and MTL areas in rats using context to guide spatial paths (project 5) and object associations (Project 4), we will directly compare the neural mechanisms used in spatial and non-spatial context-guided associations. Project 5 thereby links the projects investigating context-guided item associations in humans (Projects 1 & 2), monkeys (Project 3), and rats (Project 4), to the enormous literature on spatial processing by the MTL structures. The specific aims for this project are:

Specific Aim 1 will investigate PFC and MTL coding during context guided route learning and memory by recording multiple single units and local field potentials (LFP). Pairs of tetrode bundles will be placed in the PFC (PL, IL, or OFC) and MTL (CA1, CA3, medial entorhinal cortex, or perirhinal cortex) to record PFC and MTL activity simultaneously. Neural coding during learning and memory performance will be analyzed in collaboration with the Analysis Core using spatial and event related statistics. Cross correlations will assess spike timing within and between structures (Shapiro & Ferbinteanu, 2006), LFP spectra, comodulation, and spike wave coherence analyses will assess network dynamics within and between pairs of MTL and PFC areas (Shirvalkar et al., 2010; Young & Shapiro, 2010).

Specific Aim 2 will determine the causal role of MTL and PFC regions in context-guided memory using bilateral or crossed ipsilateral temporary inactivation of the same region pairs as described in Aim 1. Local circuits will reversibly inactivated during learning, performance, and probe tests to analyze the contribution of different PFC and MTL subregions to context-guided journeys.

Specific Aim 3 will determine the causal interactions of MTL-PFC processing interactions by combining local circuit inactivation and electrophysiology. Inactivating circuits in one region while recording in another will determine coding dependencies, testing directly how memory related firing patterns in the MTL and PFC depend upon and influence one another during context guided journeys.

Project 6

Computational models of PFC-MTL interactions.

Michael Hasselmo

Boston University


Overview. Experimental data from Projects 1 to 5 will converge on updating a single computational model of the physiological interactions of PFC and MTL that mediate context-based responses in the behavioral tasks.  The model will provide a unified theory of physiological mechanisms for representing context that links the different levels of data, allowing generation of predictions that can be compared with data to analyze and update the model.

Specific Aim 1 Modeling rodent spiking and field potentials during context-based behavior.  Modeling of PFC-MTL interactions will use different physiological mechanisms for representing context and reinforcement value for gating behavioral responses. Patterns of neural activity predicted by different physiological mechanisms for representing context and reinforcement value will guide analysis of prefrontal, parahippocampal and hippocampal units recorded in both Project 4 and 5.

Specific Aim 2 Modeling primate spiking and regional activation during context-based behavior.  The same shared model of physiological mechanisms for PFC-MTL interactions used for rodents will be applied to the task used in primates. For both Project 2 and 3, modeling will generate and test predictions about the activity and functional connectivity of PFC and MTL regions during the inference of context-based responses to cues presented in novel locations.

Specific Aim 3  Changes in neural activation and behavior with inactivation of different regions. Using the same physiological mechanisms from the shared model, we will simulate the effect of inactivation of specific subregions on preferential visual exploration and explicit choice performance in Project 1 and the neural activity in other parts of the model compared with unit recording during inactivation in Project 4 and Project 5.


Website/Resource-Sharing/Dissemination Plan. We will promote data dissemination by web posting of access to brain imaging and neurophysiology data and computer simulations and software tools, as well as research papers.

  1. Hasselmo’s neural simulator, Catacomb (, modeling scripts developed within the MATLAB simulator.  These software packages are freely available, have open source code, and are used by a number of laboratories worldwide.