Phase locking of hippocampal CA3 neurons to distal CA1 theta oscillations selectively predicts memory performance.

How the coordination of neuronal spiking and brain rhythms between hippocampal subregions supports memory function remains elusive. We studied the interregional coordination of CA3 neuronal spiking with CA1 theta oscillations by recording electrophysiological signals along the proximodistal axis of the hippocampus in rats that were performing a high-memory-demand recognition memory task adapted from humans. We found that CA3 population spiking occurs preferentially at the peak of distal CA1 theta oscillations when memory was tested but only when previously encountered stimuli were presented. In addition, decoding analyses revealed that only population cell firing of proximal CA3 together with that of distal CA1 can predict performance at test in the present non-spatial task. Overall, our work demonstrates an important role for the synchronization of CA3 neuronal activity with CA1 theta oscillations during memory testing.

Recalling gist memory depends on CA1 hippocampal neurons for lifetime retention and CA3 neurons for memory precision.

Why some of us remember events more clearly than others and why memory loses precision over time is a major focus in memory research. Here, we show that the recruitment of specific neuroanatomical pathways within the medial temporal lobe (MTL) of the brain defines the precision of the memory recalled over the lifespan. Using optogenetics, neuronal activity mapping, and studying recent to very remote memories, we report that the hippocampal subfield CA1 is necessary for retrieving the gist of events and receives maximal support from MTL cortical areas (MEC, LEC, PER, and POR) for recalling the most remote memories. In contrast, reduction of CA3's activity alone coincides with the loss of memory precision over time. We propose that a shift between specific MTL subnetworks over time might be a fundamental mechanism of memory consolidation.

Brief neuronal afterdischarges in the rat hippocampus lead to transient changes in oscillatory activity and to a very long-lasting decline in BOLD signals without inducing a hypoxic state.

The effects of hippocampal neuronal afterdischarges (nAD) on hemodynamic parameters, such as blood-oxygen-level-dependent (BOLD) signals) and local cerebral blood volume (CBV) changes, as well as neuronal activity and metabolic parameters in the dentate gyrus, was investigated in rats by combining in vivo electrophysiology with functional magnetic resonance imaging (fMRI) or 1H-nuclear magnetic resonance spectroscopy (1H-NMRS). Brief electrical high-frequency pulse-burst stimulation of the right perforant pathway triggered nAD, a seizure-like activity, in the right dentate gyrus with a high incidence, a phenomenon that in turn caused a sustained decrease in BOLD signals for more than 30 min. The decrease was associated with a reduction in CBV but not with signs of hypoxic metabolism. nAD also triggered transient changes mainly in the low gamma frequency band that recovered within 20 min, so that the longer-lasting altered hemodynamics reflected a switch in blood supply rather than transient changes in ongoing neuronal activity. Even in the presence of reduced baseline BOLD signals, neurovascular coupling mechanisms remained intact, making long-lasting vasospasm unlikely. Subsequently generated nAD did not further alter the baseline BOLD signals. Similarly, nAD did not alter baseline BOLD signals when acetaminophen was previously administered, because acetaminophen alone had already caused a similar decrease in baseline BOLD signals as observed after the first nAD. Thus, at least two different blood supply states exist for the hippocampus, one low and one high, with both states allowing similar neuronal activity. Both acetaminophen and nAD switch from the high to the low blood supply state. As a result, the hemodynamic response function to an identical stimulus differed after nAD or acetaminophen, although the triggered neuronal activity was similar.

The contributions of entorhinal cortex and hippocampus to error driven learning.

Computational models proposed that the medial temporal lobe (MTL) contributes importantly to error-driven learning, though little direct in-vivo evidence for this hypothesis exists. To test this, we recorded in the entorhinal cortex (EC) and hippocampus (HPC) as macaques performed an associative learning task using an error-driven learning strategy, defined as better performance after error relative to correct trials. Error-detection signals were more prominent in the EC relative to HPC. Early in learning hippocampal but not EC neurons signaled error-driven learning by increasing their population stimulus-selectivity following error trials. This same pattern was not seen in another task where error-driven learning was not used. After learning, different populations of cells in both the EC and HPC signaled long-term memory of newly learned associations with enhanced stimulus-selective responses. These results suggest prominent but differential contributions of EC and HPC to learning from errors and a particularly important role of the EC in error-detection.

Regional Specific Evidence for Memory-Load Dependent Activity in the Dorsal Subiculum and the Lateral Entorhinal Cortex.

The subiculum and the lateral entorhinal cortex (LEC) are the main output areas of the hippocampus which contribute to spatial and non-spatial memory. The proximal part of the subiculum (bordering CA1) receives heavy projections from the perirhinal cortex and the distal part of CA1 (bordering the subiculum), both known for their ties to object recognition memory. However, the extent to which the proximal subiculum contributes to non-spatial memory is still unclear. Comparatively, the involvement of the LEC in non-spatial information processing is quite well known. However, very few studies have investigated its role within the frame of memory function. Thus, it is not known whether its contribution depends on memory load. In addition, the deep layers of the EC have been shown to be predictive of subsequent memory performance, but not its superficial layers. Hence, here we tested the extent to which the proximal part of the subiculum and the superficial and deep layers of the LEC contribute to non-spatial memory, and whether this contribution depends on the memory load of the task. To do so, we imaged brain activity at cellular resolution in these areas in rats performing a delayed nonmatch to sample task based on odors with two different memory loads (5 or 10 odors). This imaging technique is based on the detection of the RNA of the immediate-early gene Arc, which is especially tied to synaptic plasticity and behavioral demands, and is commonly used to map activity in the medial temporal lobe. We report for the first time that the proximal part of the subiculum is recruited in a memory-load dependent manner and the deep layers of the LEC engaged under high memory load conditions during the retrieval of non-spatial memory, thus shedding light on the specific networks contributing to non-spatial memory retrieval.

fMRI of the face-processing network in the ventral temporal lobe of awake and anesthetized macaques.

The primate brain features specialized areas devoted to processing of faces, which human imaging studies localized in the superior temporal sulcus (STS) and ventral temporal cortex. Studies in macaque monkeys, in contrast, revealed face selectivity predominantly in the STS. While this discrepancy could result from a true species difference, it may simply be the consequence of technical difficulties in obtaining high-quality MR images from the ventral temporal lobe. By using an optimized fMRI protocol we here report face-selective areas in ventral TE, the parahippocampal cortex, the entorhinal cortex, and the hippocampus of awake macaques, in addition to those already known in the STS. Notably, the face-selective activation of these memory-related areas was observed although the animals were passively viewing and it was preserved even under anesthesia. These results point to similarly extensive cortical networks for face processing in humans and monkeys and highlight potential homologs of the human fusiform face area.

Frontoparietal activity with minimal decision and control in the awake macaque at 7 T.

Previous imaging work has identified a frontoparietal network in the human brain involved in many different cognitive functions, as well as in simple updates of attended information. To determine whether a similar network is present in the monkey brain and direct future electrophysiological recordings, we examined the activation of frontoparietal areas during visual stimulation in the awake, fixating monkey. We measured activity with BOLD fMRI in three animals and analyzed the data individually for each animal and at group level. We found reliable activations in lateral prefrontal and parietal areas, even though task-related decision making was minimal, as a response to simple update of visual information. These activations were significant for each individual animal, as well as at group level. Similar to human imaging results the update of visual input was enough to activate an extensive network of frontoparietal cortex in the macaque brain, a network which is normally associated with complex cognitive control processes.

Comparison of pattern recognition methods in classifying high-resolution BOLD signals obtained at high magnetic field in monkeys.

Pattern recognition methods have shown that functional magnetic resonance imaging (fMRI) data can reveal significant information about brain activity. For example, in the debate of how object categories are represented in the brain, multivariate analysis has been used to provide evidence of a distributed encoding scheme [Science 293:5539 (2001) 2425-2430]. Many follow-up studies have employed different methods to analyze human fMRI data with varying degrees of success [Nature reviews 7:7 (2006) 523-534]. In this study, we compare four popular pattern recognition methods: correlation analysis, support-vector machines (SVM), linear discriminant analysis (LDA) and Gaussian naïve Bayes (GNB), using data collected at high field (7 Tesla) with higher resolution than usual fMRI studies. We investigate prediction performance on single trials and for averages across varying numbers of stimulus presentations. The performance of the various algorithms depends on the nature of the brain activity being categorized: for several tasks, many of the methods work well, whereas for others, no method performs above chance level. An important factor in overall classification performance is careful preprocessing of the data, including dimensionality reduction, voxel selection and outlier elimination.

fMRI of the temporal lobe of the awake monkey at 7 T.

Increasingly 7 T scanners are used for fMRI of humans and non-human primates, promising improvements in signal-to-noise, spatial resolution and specificity. A disadvantage of fMRI at 7 T, but already at 3 T, is that susceptibility artifacts from air-filled cavities like the ear canal and nasal cavity cause signal loss and distortion. This limits the applicability of fMRI in these areas, thereby limiting study of these areas, but it also limits study of processes that span large-scale cortical networks or the entire brain. Our goal is to study the inferior temporal (IT) lobe in awake monkeys because of its importance in object perception and recognition, but the functional signal is degraded by strong susceptibility gradients. To allow fMRI of this region, we used an optimized SE-EPI, which recovers signal lost with GE-EPI and we corrected for susceptibility-induced image distortion. SE-EPI has the added advantage that, in contrast to GE-EPI, where the functional signal derives to a large extent from veins, the SE-EPI signal arises from the microvasculature, and hence it better represents the neural activation. We show fMRI at 7 T of the entire visual pathway in the awake primate with robust and widespread activation in all ventral areas of the brain, including areas adjacent to the ear canal. This allows fMRI of areas that normally suffer from artifact and thus more reliable whole-brain studies.

Robust controlled functional MRI in alert monkeys at high magnetic field: effects of jaw and body movements.

The use of functional magnetic resonance imaging (fMRI) in alert non-human primates is of great potential for research in systems neuroscience. It can be combined with invasive techniques and afford better understanding of non-invasively acquired brain imaging signals in humans. However, the difficulties in optimal application of alert monkey fMRI are multi-faceted, especially at high magnetic fields where the effects of motion and of changes in B0 are greatly amplified. To overcome these difficulties, strict behavioral controls and elaborate animal-training are needed. Here, we introduce a number of hardware developments, quantify the effect of movements on fMRI data, and present procedures for animal training and scanning for well-controlled and artifact-reduced alert monkey fMRI at high magnetic field. In particular, we describe systems for monitoring jaw and body movements, and for accurately tracking eye movements. A link between body and jaw movement and MRI image artifacts is established, showing that relying on the immobilization of an animal's head is not sufficient for high-quality imaging. Quantitative analysis showed that body and jaw movement events caused large instabilities in fMRI time series. On average, body movement events caused larger instabilities than jaw movement events. Residual baseline brain image position and signal amplitude shifts were observed after the jaw and body movement events ended. Based on these findings, we introduce a novel behavioral paradigm that relies on training the monkeys to stay still during long trials. A corresponding analysis method discards all data that were not obtained during the movement-free periods. The baseline position and amplitude shifts are overcome by motion correction and trial-by-trial signal normalization. The advantages of the presented method over conventional scanning and analysis are demonstrated with data obtained at 7 T. It is anticipated that the techniques presented here will prove useful for alert monkey fMRI at any magnetic field.

A novel functional magnetic resonance imaging compatible search-coil eye-tracking system.

Measuring eye movements (EMs) using the search-coil eye-tracking technique is superior to video-based infrared methods [Collewijn H, van der Mark F, Jansen TC. Precise recording of human eye movements. Vision Res 1975;15(3):447-50], which suffer from the instability of pupil size, blinking behavior and lower temporal resolution. However, no conventional functional magnetic resonance imaging (fMRI)-compatible search-coil eye tracker exists. The main problems for such a technique are the interaction between the transmitter coils and the magnetic gradients used for imaging as well as the limited amount of space in a scanner. Here we present an approach to overcome these problems and we demonstrate a method to record EMs in an MRI scanner using a search coil. The system described has a spatial resolution of 0.07 degrees (visual angle) and a high temporal resolution (22 kHz). The transmitter coils are integrated into the visual presentation system and the control/analysis unit is portable, which enables us to integrate the eye tracker with an MRI scanner. Our tests demonstrate low noise in the recorded eye traces and scanning with minimal artifact. Furthermore, the induced current in the search coil caused by the RF pulses does not lead to measurable heating. Altogether, this MR-compatible search-coil eye tracker can be used to precisely monitor EMs with high spatial and temporal resolution during fMRI. It can therefore be of great importance for studies requiring accurate fixation of a target, or measurement and study of the subject's oculomotor system.

Bayesian integration in force estimation.

When we interact with objects in the world, the forces we exert are finely tuned to the dynamics of the situation. As our sensors do not provide perfect knowledge about the environment, a key problem is how to estimate the appropriate forces. Two sources of information can be used to generate such an estimate: sensory inputs about the object and knowledge about previously experienced objects, termed prior information. Bayesian integration defines the way in which these two sources of information should be combined to produce an optimal estimate. To investigate whether subjects use such a strategy in force estimation, we designed a novel sensorimotor estimation task. We controlled the distribution of forces experienced over the course of an experiment thereby defining the prior. We show that subjects integrate sensory information with their prior experience to generate an estimate. Moreover, subjects could learn different prior distributions. These results suggest that the CNS uses Bayesian models when estimating force requirements.