Spatial memory encoding is associated with the anterior and posterior hippocampus: An fMRI activation likelihood estimation meta-analysis.

It has been hypothesized that differential processing occurs along the longitudinal (anterior-posterior) axis of the hippocampus. One hypothesis is that spatial memory (during both encoding and retrieval) is associated with the posterior hippocampus. An alternative hypothesis is that memory encoding (either spatial or nonspatial) is associated with the anterior hippocampus and memory retrieval is associated with the posterior hippocampus. Of importance, during spatial memory encoding, the spatial-posterior hypothesis predicts posterior hippocampal involvement, whereas the encoding-retrieval hypothesis predicts anterior hippocampal involvement. To distinguish between these hypotheses, we conducted a coordinate-based fMRI activation likelihood estimation (ALE) meta-analysis of 26 studies (with a total of 435 participants) that reported hippocampal activity during spatial memory encoding and/or spatial memory retrieval. Both spatial memory encoding and spatial memory retrieval produced extensive activity along the longitudinal axis of the hippocampus as well as the entorhinal cortex, the perirhinal cortex, and the parahippocampal cortex. Critically, the contrast of spatial memory encoding and spatial memory retrieval produced activations in both the anterior hippocampus and the posterior hippocampus. That spatial memory encoding produced activity in both the anterior and posterior hippocampus can be taken to reject strict forms of the spatial-posterior hypothesis, which stipulates that all forms of spatial memory produce activity in the posterior hippocampus, and the encoding-retrieval hypothesis, which stipulates that all forms of encoding versus retrieval produce activity in only the anterior hippocampus. Our results indicate that spatial memory encoding can involve the anterior hippocampus and the posterior hippocampus.

True and false memories for spatial location evoke more similar patterns of brain activity in males than females.

True and false memories recruit a number of shared brain regions; however, they are not completely overlapping. Extensive sex differences have been identified in the brain during true memories and, recently, we identified sex differences in the brain during false memories. In the current fMRI study, we sought to determine whether sex differences existed in the location and extent of overlap between true and false memories. True and false memories activated a number of shared brain regions. Compared to females, males produced a greater number of overlapping brain regions (8 versus 2 activations for males and females, respectively) including the prefrontal cortex, parietal cortex, and early/late visual processing cortices (including V1) in males and prefrontal and parietal cortices in females. Males had significantly higher similarity between true and false memory activation maps, revealed by our novel multi-voxel pattern correlation analysis. Moreover, higher similarity between true and false memory activation maps was associated with higher false memory rates. The current results suggest that true and false memories are more similar in males than females. The significant brain-behavior relationship also suggests that males may be more susceptible to false memory errors due to their highly similar true memory-false memory cortical representations.

Impaired cognitive performance in older adults is associated with deficits in item memory and memory for object features.

Amnestic mild cognitive impairment (aMCI) is associated with damage to the perirhinal/entorhinal cortex, and consequently, deficits in item/object memory. However, cognitive assessments commonly used to identify individuals with aMCI require a clinician to administer and interpret the test. We developed a novel self-administered global cognitive assessment, called the Cognitive Assessment via Keyboard (CAKe). To assess the relationship between CAKe performance and perirhinal/entorhinal cortex-dependent memory function, participants completed the CAKe, a feature source memory task, and a context memory task. During the memory tasks, participants studied line drawings with either a green or orange internal color (feature memory runs) or external color (context memory runs) and then classified each item as old and previously presented with a "green" or "orange" color, or "new". CAKe scores were correlated with item memory accuracies and source memory accuracies on both tasks. Participants with 'impaired' CAKe performance had worse item memory and worse feature source memory accuracies than those with 'normal' CAKe performance. These results demonstrate specific deficits in item memory and feature source memory and suggest that our assessments may be a valid predictor of aMCI memory deficits.

Sex is predicted by spatial memory multivariate activation patterns.

Whether sex differences exist in the brain at the macroscopic level, as measured with magnetic resonance imaging (MRI), is a topic of debate. The present spatial long-term memory functional MRI (fMRI) study predicted sex based on event-related patterns of brain activity. Within spatial memory regions of interest, patterns of activity associated with females and males were used to predict the sex of each member of left-out female-male pairs at above-chance accuracy. The current results provide evidence for sex differences in the brain processes underlying spatial long-term memory. This is the first time that sex has been predicted using event-related fMRI activation patterns. The present findings contribute to a growing body of evidence that there are functional and anatomic sex differences in the brain and, more broadly, question the widespread practice of collapsing across sex in the field of cognitive neuroscience.

Distinct patterns of hippocampal activity associated with color and spatial source memory.

The hippocampus is known to be involved in source memory across a wide variety of stimuli and source types. Thus, source memory activity in the hippocampus is thought to be domain-general such that different types of source information are similarly processed in the hippocampus. However, there is some evidence of domain-specificity for spatial and temporal source information. The current fMRI study aimed to determine whether patterns of activity in the hippocampus differed for two types of visual source information: spatial location and background color. Participants completed three runs of a spatial memory task and three runs of a color memory task. During the study phase, 32 line drawings of common objects and animals were presented to either the left or right of fixation for the spatial memory task or on either a red or green background for the color memory task. During the test phase of both tasks, 48 object word labels were presented in the center of the screen and participants classified the corresponding item as old and previously on the "left"/on a "green" background, old and previously on the "right"/on a "red" background, or "new." Two analysis methods were employed to assess whether hippocampal activity differed between the two source types: a general linear model analysis and a classification-based searchlight multivoxel pattern analysis (MVPA). The searchlight MVPA revealed that activity associated with spatial memory and color memory could be classified with above-chance accuracy in a region of the right anterior hippocampus, and a follow-up analysis revealed that there was a significant effect of memory accuracy. These results indicate that different types of source memory are represented by distinct patterns of activity in the hippocampus.

Sex differences in hippocampal connectivity during spatial long-term memory.

Sex differences in brain activity have been reported across various types of long-term memory. To our knowledge, sex differences in functional connectivity during long-term memory have not been investigated. A previous study on the structural connectome identified that female brains have a greater degree of interhemispheric connectivity than males, whereas males have a greater degree of intrahemispheric connectivity than females. The aim of the current investigation was twofold: (a) identify which brain regions were functionally connected to the hippocampus during spatial long-term memory, and (b) determine if there were sex differences in the functionally connected regions. During the study phase, abstract shapes were presented to the left or right of fixation. During the test phase, abstract shapes were presented at fixation and participants classified each item as previously on the "left" or "right". A hippocampal region of interest (ROI) was identified by contrasting spatial memory hits and misses. The peak coordinate from this ROI was used to define the center of a sphere that was used as the seed for the functional connectivity analysis. The connectivity analysis produced many connected activations including the medial posterior frontal cortex, lateral posterior frontal cortex, left inferior frontal gyrus, posterior cingulate cortex, and caudate/putamen. Although there were no regions with greater connectivity in females than males, the male versus female comparison produced connected activations in the medial posterior frontal cortex, anterior prefrontal cortex, precuneus, and cingulate sulcus. Females also had greater interhemispheric connectivity than males. The current findings suggest collapsing across sex in cognitive neuroscience studies may not be warranted.

Functional connectivity with the anterior and posterior hippocampus during spatial memory.

Evidence of differential connectivity and activity patterns across the long-axis of the hippocampus has led to many hypotheses about functional specialization of the anterior and posterior hippocampus, including a hypothesis linking the anterior hippocampus to memory encoding and the posterior hippocampus to memory retrieval. The hippocampal encoding/retrieval and network (HERNET) model of memory predicts that encoding should engage the anterior hippocampus and the attention network, whereas retrieval should engage the posterior hippocampus and the default network. In a previous fMRI study that employed multivoxel pattern analysis, we found that the patterns of activity in the anterior hippocampus predicted the quadrant of spatial memory encoding. In the current fMRI study, we investigated whether the spatial memory encoding activity in the anterior hippocampus and retrieval activity in the posterior hippocampus had a higher degree of connectivity to the attention network or the default network. During the study phase, abstract shapes were presented in each quadrant of the visual field and participants were instructed to remember each shape's location while maintaining central fixation. During the test phase, the same shapes were presented in the center of the screen and participants identified the previous location of each shape. Generalized psychophysiological interaction analyses were conducted between the anatomically defined anterior and posterior hippocampus and the rest of the brain. This revealed preferential connectivity between the anterior hippocampus and regions of the attention network during encoding and between the posterior hippocampus and regions of the default network during retrieval. In addition, there were location-specific patterns of connectivity with the anterior hippocampus and posterior hippocampus during encoding and retrieval of right visual field items. These results suggest that the anterior and posterior hippocampus interact with regions of the attention network and default network during spatial memory encoding and retrieval, respectively, and support the HERNET model of memory.

Long-term memory specificity depends on inhibition of related items.

Long-term memory relies on both accurately retrieving specific details and inhibiting competing information. In the current investigation, we evaluated the specificity of long-term memory representations for faces. During each study phase, participants were presented with neutral Caucasian male and female faces. During the corresponding test phase, old faces, related faces, and new faces were presented and participants made "old"-"new" recognition judgments. Related faces were created by morphing along a continuum in steps of 20% (i.e., 20%, 40%, 60% and 80% morphs) between old faces and new faces (independent ratings indicated that the pairs of to-be-morphed old faces and new faces were perceptually dissimilar). In two experiments, memory representations were very specific as the "old" response rate for old faces was significantly higher than closely related faces (i.e., 20% morphs). Furthermore, there was evidence of memory inhibition, as the "old" response rate for 20% morphs was significantly lower than 40% morphs (the identical pattern of results was observed with a d' analysis). These findings may reflect an evolutionary advantage for recognising specific faces, which may require inhibition of closely related faces.

Different patterns of cortical activity in females and males during spatial long-term memory.

It is generally assumed that identical neural regions mediate the same cognitive functions in females and males. However, anatomic and molecular sex differences exist in the brain, including in regions associated with long-term memory, which suggests there may be functional differences. The present functional magnetic resonance imaging (fMRI) investigation aimed to identify the differences and similarities in brain activity between females and males during spatial long-term memory. During encoding, abstract shapes were presented to the left or right of fixation. During retrieval, shapes were presented at fixation and participants made "old-left" or "old-right" judgments. For both females and males, spatial memory hits versus misses produced activity in regions commonly associated with visual long-term memory; however, the activations were almost completely distinct between the sexes. An interaction analysis revealed sex-specific activity for males in visual processing regions, the left putamen, the right caudate nucleus, and bilateral cerebellum, and sex-specific activity for females in the parietal cortex. A targeted anatomic region-of-interest (ROI) analysis identified sex-specific activity for males and females in the left hippocampus and language processing cortex, respectively. A multi-voxel pattern correlation analysis within functional ROIs between all pairs of participants showed greater within-sex than between-sex correlations, indicating the differential activations were due to sex differences rather than other individual differences between groups. These results indicate that spatial long-term memory is mediated by largely different brain regions in females and males. These findings have major implications for the field of cognitive neuroscience, where it is common practice to collapse across sex.

Similar patterns of cortical activity in females and males during item memory.

Anatomic and molecular sex differences exist in the brain, which suggests there may be functional differences. The present functional magnetic resonance imaging (fMRI) investigation aimed to identify the similarities and differences in brain activity between females and males during item memory. During encoding, abstract shapes were presented to the left or right of fixation. During retrieval, old and new shapes were presented at fixation and participants made "old-left", "old-right", or "new" judgments. Item memory was isolated by contrasting correct "old" responses to old items (with incorrect spatial memory responses; item memory hits) and "new" responses to old items (item memory misses). For both sexes, item memory produced activity in regions associated with visual long-term memory including the prefrontal cortex, parietal cortex, and visual processing regions. A sex by accuracy interaction analysis within each sub-region of activity produced largely null results, supporting common patterns of brain activity. However, there was sex-specific (male > female) activity within default network regions, which suggests males may have been less engaged in the task, and there was evidence for greater activity for females than males in language processing cortex. The present findings indicate that females and males employ similar patterns of brain activity during item memory.

False memories for shape activate the lateral occipital complex.

Previous functional magnetic resonance imaging evidence has shown that false memories arise from higher-level conscious processing regions rather than lower-level sensory processing regions. In the present study, we assessed whether the lateral occipital complex (LOC)-a lower-level conscious shape processing region-was associated with false memories for shape. During encoding, participants viewed intact or scrambled colored abstract shapes. During retrieval, colored disks were presented and participants indicated whether the corresponding item was previously "intact" or "scrambled." False memories for shape ("intact"/scrambled > "scrambled"/scrambled) activated LOC, which indicates lower-level sensory processing regions can support false memory.

An attention account of neural priming.

Repetition priming of familiar stimuli (e.g., objects) produces a decrease in visual cortical activity for repeated versus novel items, which has been attributed to more fluent processing for repeated items. By contrast, priming of unfamiliar stimuli (e.g., abstract shapes) produces an increase in visual cortical activity. The mechanism for priming-related increases in activity for repeated unfamiliar stimuli is unknown. We hypothesised that such increases in activity may reflect attentional allocation to these items. We tested this hypothesis using a priming-spatial attention paradigm. During Phase 1 of Experiment 1, participants viewed unfamiliar abstract shapes and familiar objects. During Phase 2, participants identified target letters (S or H). Each target letter was preceded by a non-informative shape or object cue that was repeated (from Phase 1) or novel in the same (valid) or opposite (invalid) hemifield. In Experiment 2, we manipulated shape familiarity by presenting shapes once or six times during Phase 1. For both experiments, at valid locations, target identification accuracy was higher following repeated versus novel unfamiliar item cues and lower following repeated versus novel familiar item cues. These findings support our hypothesis that priming-related increases in visual cortical activity for repeated unfamiliar items may, in part, reflect attentional allocation.

Repetition Enhancement of Amygdala and Visual Cortex Functional Connectivity Reflects Nonconscious Memory for Negative Visual Stimuli.

Most studies using a recognition memory paradigm examine the neural processes that support the ability to consciously recognize past events. However, there can also be nonconscious influences from the prior study episode that reflect repetition suppression effects-a reduction in the magnitude of activity for repeated presentations of stimuli-that are revealed by comparing neural activity associated with forgotten items to correctly rejected novel items. The present fMRI study examined the effect of emotional valence (positive vs. negative) on repetition suppression effects. Using a standard recognition memory task, 24 participants viewed line drawings of previously studied negative, positive, and neutral photos intermixed with novel line drawings. For each item, participants made an old-new recognition judgment and a sure-unsure confidence rating. Collapsed across valence, repetition suppression effects were found in ventral occipital-temporal cortex and frontal regions. Activity levels in the majority of these regions were not modulated by valence. However, repetition enhancement of the amygdala and ventral occipital-temporal cortex functional connectivity reflected nonconscious memory for negative items. In this study, valence had little effect on activation patterns but had a larger effect on functional connectivity patterns that were markers of nonconscious memory. Beyond memory and emotion, these findings are relevant to other cognitive and social neuroscientists that utilize fMRI repetition effects to investigate perception, attention, social cognition, and other forms of learning and memory.

Recollection is a continuous process: Evidence from plurality memory receiver operating characteristics.

Is recollection a continuous/graded process or a threshold/all-or-none process? Receiver operating characteristic (ROC) analysis can answer this question as the continuous model and the threshold model predict curved and linear recollection ROCs, respectively. As memory for plurality, an item's previous singular or plural form, is assumed to rely on recollection, the nature of recollection can be investigated by evaluating plurality memory ROCs. The present study consisted of four experiments. During encoding, words (singular or plural) or objects (single/singular or duplicate/plural) were presented. During retrieval, old items with the same plurality or different plurality were presented. For each item, participants made a confidence rating ranging from "very sure old", which was correct for same plurality items, to "very sure new", which was correct for different plurality items. Each plurality memory ROC was the proportion of same versus different plurality items classified as "old" (i.e., hits versus false alarms). Chi-squared analysis revealed that all of the plurality memory ROCs were adequately fit by the continuous unequal variance model, whereas none of the ROCs were adequately fit by the two-high threshold model. These plurality memory ROC results indicate recollection is a continuous process, which complements previous source memory and associative memory ROC findings.

The hippocampus and implicit memory.

The traditional memory-systems view is that explicit (conscious) long-term memory is associated with the hippocampus and implicit (nonconscious) memory is associated with non-hippocampal brain regions. This special issue of Cognitive Neuroscience focuses on whether the hippocampus is associated with implicit memory. An empirical fMRI paper by Miller, Kennard, Gowland, Antoniades, and Rosenthal (this issue) evaluated recognition memory performance of autobiographical amnesia patients with bilateral damage to hippocampal sub-region CA3 and found they had greater than chance recognition memory performance for spatial sequence learning, spatial item learning-same location, and non-spatial item learning, but chance performance for non-spatial sequence learning and spatial item learning-different location. These results are at odds with the view that the hippocampus is generally involved in sequence learning and complex event recognition. A discussion paper by Steinkrauss and Slotnick (this issue) considered whether fMRI studies have provided evidence that the hippocampus is associated with implicit memory. It was argued that all previous studies have been confounded by explicit memory, attentional states, stimuli, or novelty. This discussion paper is followed by commentaries from Hannula (this issue), Henke and Ruch (this issue), Rosenthal (this issue), Spaak (this issue), Thakral et al. (this issue), and Züst (this issue). The articles in this special issue illustrate that the association between the hippocampus and implicit memory is under active investigation and debate. It is hoped that the evidence and discourse in this issue will provide directions for future research.

Is implicit memory associated with the hippocampus?

According to the traditional memory-systems view, the hippocampus is critical during explicit (conscious) long-term memory, whereas other brain regions support implicit (nonconscious) memory. In the last two decades, some fMRI studies have reported hippocampal activity during implicit memory tasks. The aim of the present discussion paper was to identify whether any implicit memory fMRI studies have provided convincing evidence that the hippocampus is associated with nonconscious processes without being confounded by conscious processes. Experimental protocol and analysis parameters included the stimulus type(s), task(s), measures of subjective awareness, explicit memory accuracy, the relevant fMRI contrast(s) or analysis, and confound(s). A systematic review was conducted to identify implicit memory studies that reported fMRI activity in the hippocampus. After applying exclusion criteria, 13 articles remained for analysis. We found that there were no implicit memory fMRI studies where subjective awareness was absent, explicit memory performance was at chance, and there were no confounds that could have driven the observed hippocampal activity. The confounds included explicit memory (including false memory), imbalanced attentional states between conditions (yielding activation of the default-mode network), imbalanced stimuli between conditions, and differential novelty. As such, not a single fMRI study provided convincing evidence that implicit memory was associated with the hippocampus. Neuropsychological evidence was also considered, and implicit memory deficits were caused by factors known to disrupt brain regions beyond the hippocampus, such that the behavioral effects could not be attributed to this region. The present results indicate that implicit memory is not associated with the hippocampus.

Can Residents With Late-Stage Dementia Still Engage?

Institutionalized persons with dementia often lack access to meaningful activity, which can lead to agitation, loneliness, and depression. Engagement in activity may improve negative symptoms but is difficult in most settings. In this study, we investigated the degree to which the Reading Buddies Program, in which occupational therapy graduate students read books with residents with dementia, engaged residents. We further assessed whether the level of engagement was affected by various parameters, including those related to interaction, environment, attention, attitude, and activity. The primary outcome measure was engagement percentage-duration of time the book was read divided by duration of time the person with dementia engaged with the book. As expected, increased attention, attitude, and activity parameters were associated with increased engagement. None of the environmental parameters significantly affected engagement. Overall, we found that reading with persons with dementia led to a very high level of engagement and appeared to reduce negative symptoms.

The fusiform face area responds equivalently to faces and abstract shapes in the left and central visual fields.

The fusiform face area (FFA) is widely believed to be specialized for processing faces. Although the FFA is most responsive to faces, this region also consistently responds to non-face items. This suggests that the FFA may be tuned to a feature that is shared by faces and non-face items. Based on the known left visual field face-processing bias along with evidence that the FFA responds to the visual feature of shape, we hypothesized that the FFA may be particularly tuned to shapes presented in the left visual field. We tested this hypothesis using functional magnetic resonance imaging (fMRI). In a face localizer run, participants viewed blocks of faces or objects. In a separate run, blocks of intact or scrambled abstract shapes were presented in the left, the central, or the right visual field. Within each of the eleven face-processing regions-of-interest (identified by contrasting faces and objects), the magnitude of activity associated with faces was compared to the magnitude of activity associated with intact shapes. Consistent with previous results, collapsing over shape visual field location, the magnitude of activity associated with faces was greater than the magnitude of activity associated with shapes in the FFA. However, separating by shape visual field location revealed an equivalent magnitude of activity associated with faces and shapes in the FFA when shapes were presented in the left and central visual fields. These findings indicate that the FFA, rather than being specialized for holistic face processing, mediates shape processing in the left and central visual fields.

No convincing evidence the hippocampus is associated with working memory.

In a previous discussion paper , twenty-six working memory fMRI studies that reported activity in the hippocampus were systematically analyzed. None of these studies provided convincing evidence that the hippocampus was active during the late delay phase, the only period in which working memory can be isolated from long-term memory processes. Based on these results, it was concluded that working memory does not activate the hippocampus. Six commentaries on the discussion paper were received from Courtney (2022), Kessels and Bergmann (2022), Peters and Reithler (2022), Rose and Chao (2022), Stern and Hasselmo (2022), and Wood et al. (2022). Based on these commentaries, the present response paper considered whether there is evidence of sustained hippocampal activity during the working memory delay period based on depth-electrode recording, whether there are activity-silent working memory mechanisms in the hippocampus, and whether there is hippocampal lesion evidence indicating this region is important for working memory. There was no convincing electrophysiological or neuropsychological evidence that the hippocampus is associated with working memory maintenance, and activity-silent mechanisms were arguably speculative. Given that only a small fraction (approximately 5%) of working memory fMRI studies have reported hippocampal activity and lesion evidence indicates the hippocampus is not necessary for working memory, the burden of proof is on proponents of the view that the hippocampus is important for working memory to provide compelling evidence to support their position. To date, from my perspective, there is no convincing evidence that the hippocampus is associated with working memory.

TMS must not harm participants: guidelines for evaluating TMS protocol safety.

Transcranial magnetic stimulation (TMS) can modulate a targeted brain region to assess whether that region is involved in a cognitive process. When TMS is employed in cognitive neuroscience, participants are typically healthy volunteers, and the technique is described as noninvasive. However, TMS parameters can be set such that stimulation produces long-lasting effects. Critically, TMS effects that have any possibility of lasting beyond a participant's time in the lab are potentially harmful. In this editorial, evidence is considered that indicates a 20-Hz multi-day TMS protocol has long-lasting effects, and a continuous theta-burst stimulation protocol needs further testing before it is deemed noninvasive. The following guidelines are provided for TMS protocol evaluation: 1) Effects must be shown to completely dissipate before participants leave the lab by testing well beyond the expected duration. 2) Participants should complete a cognitive test battery before TMS and after the effects are expected to dissipate. 3) Protocols should not be employed that produce effects lasting longer than the time in the lab. 4) The number of participants should ensure error bars are small, and results generalize to the population. 5) Results should be assessed at the group and individual-participant level, and effects should dissipate for every participant. 6) Bayesian analysis should be conducted to evaluate evidence in favor of the null hypothesis. 7) Effects should be assessed in multiple cortical regions. It is hoped that these guidelines will be employed to ensure the continued use of TMS as a valuable tool in the field of cognitive neuroscience.