Bejjanki, Vikranth R.; da Silveira, Rava Azeredo; Cohen, Jonathan D.; Turk-Browne, Nicholas B.
Multivariate decoding methods, such as multivoxel pattern analysis (MVPA), are highly effective at extracting information from brain imaging data. Yet, the precise nature of the information that MVPA draws upon remains controversial. Most current theories emphasize the enhanced sensitivity imparted by aggregating across voxels that have mixed and weak selectivity. However, beyond the selectivity of individual voxels, neural variability is correlated across voxels, and such noise correlations may contribute importantly to accurate decoding. Indeed, a recent computational theory proposed that noise correlations enhance multivariate decoding from heterogeneous neural populations. Here we extend this theory from the scale of neurons to functional magnetic resonance imaging (fMRI) and show that noise correlations between heterogeneous populations of voxels (i.e., voxels selective for different stimulus variables) contribute to the success of MVPA. Specifically, decoding performance is enhanced when voxels with high vs. low noise correlations (measured during rest or in the background of the task) are selected during classifier training. Conversely, voxels that are strongly selective for one class in a GLM or that receive high classification weights in MVPA tend to exhibit high noise correlations with voxels selective for the other class being discriminated against. Furthermore, we use simulations to show that this is a general property of fMRI data and that selectivity and noise correlations can have distinguishable influences on decoding. Taken together, our findings demonstrate that if there is signal in the data, the resulting above-chance classification accuracy is modulated by the magnitude of noise correlations.
Rafidi, Nicole S; Hulbert, Justin C; Brooks, Paula P; Norman, Kenneth A
Repeated testing (as opposed to repeated study) leads to improved long-term memory retention, but the mechanism underlying this improvement remains controversial. In this work, we test the hypothesis that retrieval practice benefits subsequent recall by reducing competition from related memories. This hypothesis implies that the degree of reduction in competition between retrieval practice attempts should predict subsequent memory for the practiced items. To test this prediction, we collected electroencephalography (EEG) data across two sessions. In the first session, participants practiced selectively retrieving exemplars from superordinate semantic categories (high competition), as well as retrieving the names of the superordinate categories from exemplars (low competition). In the second session, participants repeatedly studied and were then tested on Swahili-English vocabulary. One week after session two, participants were again tested on the vocabulary. We trained a within-subject classifier on the data from session one to distinguish high and low competition states. We then used this classifier to measure competition across multiple retrieval practice attempts in the second session. The degree to which competition decreased for a given vocabulary word predicted whether that item was subsequently remembered in the third session. These results are consistent with the hypothesis that repeated testing improves retention by reducing competition.
Our daily lives revolve around sharing experiences and memories with others. When different people recount the same events, how similar are their underlying neural representations? In this study, participants viewed a fifty-minute audio-visual movie, then verbally described the events while undergoing functional MRI. These descriptions were completely unguided and highly detailed, lasting for up to forty minutes. As each person spoke, event-specific spatial patterns were reinstated (movie-vs.-recall correlation) in default network, medial temporal, and high-level visual areas; moreover, individual event patterns were highly discriminable and similar between people during recollection (recall-vs.-recall similarity), suggesting the existence of spatially organized memory representations. In posterior medial cortex, medial prefrontal cortex, and angular gyrus, activity patterns during recall were more similar between people than to patterns elicited by the movie, indicating systematic reshaping of percept into memory across individuals. These results reveal striking similarity in how neural activity underlying real-life memories is organized and transformed in the brains of different people as they speak spontaneously about past events.
Antony, James W.; Piloto, Luis; Wang, Margaret; Brooks, Paula P.; Norman, Kenneth A.; Paller, Ken A.
The stability of long-term memories is enhanced by reactivation during sleep. Correlative evidence has linked memory reactivation with thalamocortical sleep spindles, although their functional role is not fully understood. Our initial study replicated this correlation and also demonstrated a novel rhythmicity to spindles, such that a spindle is more likely to occur approximately 3–6 s following a prior spindle. We leveraged this rhythmicity to test the role of spindles in memory by using real-time spindle tracking to present cues within versus just after the presumptive refractory period; as predicted, cues presented just after the refractory period led to better memory. Our findings demonstrate a precise temporal link between sleep spindles and memory reactivation. Moreover, they reveal a previously undescribed neural mechanism whereby spindles may segment sleep into two distinct substates: prime opportunities for reactivation and gaps that segregate reactivation events.
This archive contains spike trains simultaneously recorded from ganglion cells in the tiger salamander retina with a multi-electrode array while viewing a repeated natural movie clip. These data have been analyzed in previous papers, notably Puchalla et al. Neuron 2005 and Schneidman et al. Nature 2006.
Chang, Claire H. C.; Lazaridi, Christina; Yeshurun, Yaara; Norman, Kenneth A.; Hasson, Uri
This study examined how the brain dynamically updates event representations by integrating new information over multiple minutes while segregating irrelevant input. A professional writer custom-designed a narrative with two independent storylines, interleaving across minute-long segments (ABAB). In the last (C) part, characters from the two storylines meet and their shared history is revealed. Part C is designed to induce the spontaneous recall of past events, upon the recurrence of narrative motifs from A/B, and to shed new light on them. Our fMRI results showed storyline-specific neural patterns, which were reinstated (i.e. became more active) during storyline transitions. This effect increased along the processing timescale hierarchy, peaking in the default mode network. Similarly, the neural reinstatement of motifs was found during part C. Furthermore, participants showing stronger motif reinstatement performed better in integrating A/B and C events, demonstrating the role of memory reactivation in information integration over intervening irrelevant events.
In the attention schema theory, the brain constructs a model of attention, the attention schema, to aid in the endogenous control of attention. Growing behavioral evidence appears to support this proposal. However, a central question remains: does a controller of attention actually benefit by having access to an attention schema? We constructed an artificial, deep Q-learning, neural network agent that was trained to control a simple form of visuospatial attention, tracking a stimulus with its attention spotlight in order to solve a catch task. The agent was tested with and without access to an attention schema. In both conditions, the agent received sufficient information such that it should, theoretically, be able to learn the task. We found that with an attention schema present, the agent learned to control its attention spotlight and learned the catch task to a high degree of performance. Once the agent learned, if the attention schema was disabled, the agent could no longer perform effectively. If the attention schema was removed before learning began, the agent was drastically impaired at learning. The results show how the presence of even a simple attention schema provides a profound benefit to a controller of attention. We interpret these results as supporting the central argument of AST: the brain evolved an attention schema because of its practical benefit in the endogenous control of attention.
Recent advances in experimental techniques have allowed the simultaneous recordings of
populations of hundreds of neurons, fostering a debate about the nature of the collective
structure of population neural activity. Much of this debate has focused on the
empirical findings of a phase transition in the parameter space of maximum entropy
models describing the measured neural probability distributions, interpreting this phase
transition to indicate a critical tuning of the neural code. Here, we instead focus on the
possibility that this is a first-order phase transition which provides evidence that the
real neural population is in a `structured', collective state. We show that this collective
state is robust to changes in stimulus ensemble and adaptive state. We find that the
pattern of pairwise correlations between neurons has a strength that is well within the
strongly correlated regime and does not require fine tuning, suggesting that this state is
generic for populations of 100+ neurons. We find a clear correspondence between the
emergence of a phase transition, and the emergence of attractor-like structure in the
inferred energy landscape. A collective state in the neural population, in which neural
activity patterns naturally form clusters, provides a consistent interpretation for our
Extrapolation -- the ability to make inferences that go beyond the scope of one's experiences -- is a hallmark of human intelligence. By contrast, the generalization exhibited by contemporary neural network algorithms is largely limited to interpolation between data points in their training corpora. In this paper, we consider the challenge of learning representations that support extrapolation. We introduce a novel visual analogy benchmark that allows the graded evaluation of extrapolation as a function of distance from the convex domain defined by the training data. We also introduce a simple technique, context normalization, that encourages representations that emphasize the relations between objects. We find that this technique enables a significant improvement in the ability to extrapolate, considerably outperforming a number of competitive techniques.