Neural mechanisms of human memory
for events and spatial locations
Space and Memory Lab: UCL Institute of Cognitive Neuroscience & UCL Institute of Neurology
Spatial memory in MEG and iEEG
Opposing effects of emotion on memory
Rodent studies suggest that spatial memory function relies on the hippocampal
formation and is mediated by theta and gamma band oscillations, phase- and
phase- amplitude coupling. We examine MEG recordings from healthy
volunteers and iEEG recordings from epileptic patients performing a spatial
memory task to ascertain if these findings can be translated to humans.
The formation of associations between items in memory relies on hippocampaldependent mechanisms that go beyond supporting memory for a single item.
We show that negative affect can faciliate item memory through amygdaladependent processes. In contrast, negative affect can disrupt associative
memory relating to reduced activity in the hippocampus.
Methods & Behavioural Results
The Task
Recognition Memory
Participants encode the location of four visible objects within a virtual environment with distal cues
for orientation. They are subsequently cued with the image of an object, placed back in the
environment, and must retrieve the location of that object and make a response.
1.0
Our findings demonstrate a significant increase in mPFC 4-8Hz theta power during the cue period
(A) compared to a baseline period of quiet fixation. There is also a significant increase in theta
phase coupling between mPFC and right medial temporal lobe (B), and increased theta phase gamma amplitude coupling between mPFC and parietal / occipital regions (C) during this time
period. These results suggest that mPFC may co-ordinate spatial memory retrieval through
oscillatory coherence between multiple cortical regions, including the hippocampus.
A
C
B
Movement Onset - iEEG Data
Interictal Spikes and Performance
Our findings demonstrate a significant
increase in theta power on depth electrode
contacts placed in the hippocampus at
movement onset in the virtual environment,
specifically during encoding trials.
Epileptic EEG recordings exhibit interictal spike
(IIS) activity that reflects the transient, abnormal
discharge of large neural populations. We find that
the frequency with which IIS events occur on depth
electrode contacts placed in the hippocampus
during the cue period is negatively correlated with
spatial memory performance.
1.0
p = 0.07
p = 0.001
Proportion Correct
p < 0.001
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
Cue Period - MEG Data
Associative Memory
Participants encoded neutral and negative images
presented as pure neutral, mixed or pure negative
pairs. At test, participants were cued with a
neutral or negative image and required to retrieve
the paired associate originally presented with the
cue.
Cue
Neutral
Target Neutral
Neutral
Negative
Negative
Neutral
Negative
Negative
0.0
p < 0.001
p < 0.05
Neutral
Neutral
Neutral
Negative
Negative
Neutral
Negative
Negative
Recognition was enhanced for negative cues.
Associative memory was reduced for pairs that
included a negative image at encoding. In
addition, retrieval of a negative target associate
was enhanced relative to being cued with a
negative image to retrieve a neutral target
Encoding
Examination of encoding
trials, irrespective of
memory performance,
showed greater activity
in the left hippocampus
for pure neutral pairs
supporting the observed
decrease in associative
memory and disruption
of normal hippocampal
function during negative
affect.
Bilateral hippocampal activity at
encoding predicted subsequent
memory for retrieving the
correct associate.
Retrieval
Illustrative results from one patient
At retrieval, accurate recognition
of a previously seen negative
image compared to neutral
images was associated with
bilateral amygdala activity.
Associative memory hits
when cued by a neutral
image were predicted by
hippocampal activity further
supporting reduced function
by negative stimuli.
Accurate associative hits
for negative target images
were
associated
with
bilateral amygdala activity
supporting its role in
facilitating negative item
retrieval.
Pattern completion in the hippocampus
Episodic memories are thought to be retrieved by a process of pattern completion in the hippocampus. We
Retrieval
provide behavioural and fMRI evidence that multi-element events are encoded by separate cortical regions
and bound together within the hippocampus. At retrieval, the reinstatement of all event elements in the
neocortex is initiated by the hippocampus.
Behavioural Dependency
Encoding
(A) Multi-element events
were
encoded
as
pairwise associations.
Events
were
either
encoded within closedloop structures (C) or
open-loop chains (D).
Memory was tested for
all associations (B).
Dependency was calculated for each event using contingency
tables and compared to dependent and independent models.
Events encoded within a closed-loop structure showed greater
behavioural dependency. Increases in dependency positively
correlated with retrieval of the non-target associate during test.
We identified specific cortical regions associated with each class
of event element during encoding.
A greater subsequent memory effect was observed in the
hippocampus for closed- versus open- loop conditions during the
third encoding trial of each event (when the loop was closed in
closed-loop events).
Event elements were reinstated
in
the
neocortex
during
retrieval. Events encoded within
a closed-loop structure showed
greater cortical reinstatement
for the non-target element
compared to events encoded as
open-loop chains.
The difference in hippocampal BOLD response between retrieval
of closed and open loop events was correlated with activity in
neocortical regions for non-target elements, suggesting that the
hippocampus initiates cortical reinstatement.