Collective minds: social network topology shapes collective cognition

Human cognition is not solitary, it is shaped by collective learning and memory. Unlike swarms or herds, human social networks have diverse topologies, serving diverse modes of collective cognition and behaviour. Here, we review research that combines network structure with psychological and neural experiments and modelling to understand how the topology of social networks shapes collective cognition. First, we review graph-theoretical approaches to behavioural experiments on collective memory, belief propagation and problem solving. These results show that different topologies of communication networks synchronize or integrate knowledge differently, serving diverse collective goals. Second, we discuss neuroimaging studies showing that human brains encode the topology of one's larger social network and show similar neural patterns to neural patterns of our friends and community ties (e.g. when watching movies). Third, we discuss cognitive similarities between learning social and non-social topologies, e.g. in spatial and associative learning, as well as common brain regions involved in processing social and non-social topologies. Finally, we discuss recent machine learning approaches to collective communication and cooperation in multi-agent artificial networks. Combining network science with cognitive, neural and computational approaches empowers investigating how social structures shape collective cognition, which can in turn help design goal-directed social network topologies. This article is part of a discussion meeting issue 'The emergence of collective knowledge and cumulative culture in animals, humans and machines'.

The Mechanisms Underlying Interference and Inhibition: A Review of Current Behavioral and Neuroimaging Research

The memory literature has identified interference and inhibition as two major sources of forgetting. While interference is generally considered to be a passive cause of forgetting arising from exposure to additional information that impedes subsequent recall of target information, inhibition concerns a more active and goal-directed cause of forgetting that can be achieved intentionally. Over the past 25 years, our knowledge of the neural mechanisms underlying both interference-induced and inhibition-induced forgetting has expanded substantially. The present paper gives a critical overview of this research, pointing out empirical gaps in the current work and providing suggestions for future studies.

The Architecture of Human Memory: Insights from Human Single-Neuron Recordings

Deciphering the mechanisms of human memory is a central goal of neuroscience, both from the point of view of the fundamental biology of memory and for its translational relevance. Here, we review some contributions that recordings from neurons in humans implanted with electrodes for clinical purposes have made toward this goal. Recordings from the medial temporal lobe, including the hippocampus, reveal the existence of two classes of cells: those encoding highly selective and invariant representations of abstract concepts, and memory-selective cells whose activity is related to familiarity and episodic retrieval. Insights derived from observing these cells in behaving humans include that semantic representations are activated before episodic representations, that memory content and memory strength are segregated, and that the activity of both types of cells is related to subjective awareness as expected from a substrate for declarative memory. Visually selective cells can remain persistently active for several seconds, thereby revealing a cellular substrate for working memory in humans. An overarching insight is that the neural code of human memory is interpretable at the single-neuron level. Jointly, intracranial recording studies are starting to reveal aspects of the building blocks of human memory at the single-cell level. This work establishes a bridge to cellular-level work in animals on the one hand, and the extensive literature on noninvasive imaging in humans on the other hand. More broadly, this work is a step toward a detailed mechanistic understanding of human memory that is needed to develop therapies for human memory disorders.

Time-Tailoring van der Waals Heterostructures for Human Memory System Programming

The human memory system plays an indispensable role in oblivion, learning, and memorization. Implementing a memory system within electronic devices contributes an important step toward constructing a neuromorphic system that emulates advanced mental functions of the human brain. Given the complex time-tailoring requirement, integrating a human memory system into one system is a great challenge. Here, one van der Waals heterostructure with flexible time-tailoring ability is demonstrated, which can meet the high requirement of human memory system programming. By stacking volatile and nonvolatile function layers, all basic memory types, including sensory memory, short-term and long-term memory are integrated into the device and the transition between these memory types are flexible. Moreover, decision-making action and in situ result storage are also demonstrated. It is anticipated that the demonstrated time-tailoring system will support the model of a human memory system.

Different reactivation procedures enable or prevent episodic memory updating

The present study asked whether the specific method of reactivation modulates the impact of new learning on reactivated episodic memories. The study consisted of three sessions that were spaced 48 hr apart. It used an ABAC paradigm that allowed for the simultaneous assessment of retroactive interference (RI: reduced A-B recall after A-C learning) and intrusions from C into A-B recall. In Session 1, participants learned a list of paired-associates A-B. In Session 2, memory for A-B was reactivated or not and then participants either learned a second list of paired-associates A-C or completed a control task. Three different reminder conditions were compared to a no-reminder condition: a test condition, in which participants were asked to recall B in response to A, a restudy condition, in which A-B pairs were presented again for study, and a cue-word only reminder condition, in which A cues were presented in an unrelated rating task. In Session 3, recall of A-B was tested. Moderate or indirect reactivation of A-B (presentation of cue-words only) resulted in high RI effects and intrusion rates, whereas strong and direct reactivation (test and restudy) drastically reduced these effects. We suggest that direct reactivation of A-B before A-C learning strengthens memory and draws attention to list differences, thereby enhancing list segregation, and reducing memory updating.

Memory, stress, and the hippocampal hypothesis: Firefighters' recollections of the fireground

Nadel, Jacobs, and colleagues have postulated that human memory under conditions of extremely high stress is "special." In particular, episodic memories are thought to be susceptible to impairment, and possibly fragmentation, attributable to hormonally based dysfunction occurring selectively in the hippocampal system. While memory for highly salient and self-relevant events should be better than the memory for less central events, an overall nonmonotonic decrease in spatio/temporal episodic memory as stress approaches traumatic levels is posited. Testing human memory at extremely high levels of stress, however, is difficult and reports are rare. Firefighting is the most stressful civilian occupation in our society. In the present study, we asked New York City firefighters to recall everything that they could upon returning from fires they had just fought. Communications during all fires were recorded, allowing verification of actual events. Our results confirmed that recall was, indeed, impaired with increasing stress. A nonmonotonic relation was observed consistent with the posited inverted u-shaped memory-stress function. Central details about emergency situations were better recalled than were more schematic events, but both kinds of events showed the memory decrement with high stress. There was no evidence of fragmentation. Self-relevant events were recalled nearly five times better than events that were not self-relevant. These results provide confirmation that memories encoded under conditions of extremely high stress are, indeed, special and are impaired in a manner that is consistent with the Nadel/Jacobs hippocampal hypothesis.

Signal Complexity of Human Intracranial EEG Tracks Successful Associative-Memory Formation across Individuals

Memory performance is highly variable among individuals. Most studies examining human memory, however, have largely focused on the neural correlates of successful memory formation within individuals, rather than the differences among them. As such, what gives rise to this variability is poorly understood. Here, we examined intracranial EEG (iEEG) recordings captured from 43 participants (23 male) implanted with subdural electrodes for seizure monitoring as they performed a paired-associates verbal memory task. We identified three separate but related signatures of neural activity that tracked differences in successful memory formation across individuals. High-performing individuals consistently exhibited less broadband power, flatter power spectral density slopes, and greater complexity in their iEEG signals. Furthermore, within individuals across three separate time scales ranging from seconds to days, successful recall was positively associated with these same metrics. Our data therefore suggest that memory ability across individuals can be indexed by increased neural signal complexity.SIGNIFICANCE STATEMENT We show that participants whose intracranial EEG exhibits less low-frequency power, flatter power spectrums, and greater sample entropy overall are better able to memorize associations, and that the same metrics track fluctuations in memory performance across time within individuals. These metrics together signify greater neural signal complexity, which may index the brain's ability to flexibly engage with information and generate separable memory representations. Critically, the current set of results provides a unique window into the neural markers of individual differences in memory performance, which have hitherto been underexplored.

BDNF Variants May Modulate Long-Term Visual Memory Performance in a Healthy Cohort

Brain-derived neurotrophic factor (BDNF) is involved in numerous cognitive functions including learning and memory. BDNF plays an important role in synaptic plasticity in humans and rats with BDNF shown to be essential for the formation of long-term memories. We previously identified a significant association between the BDNF Val66Met polymorphism (rs6265) and long-term visual memory (p-value = 0.003) in a small cohort (n = 181) comprised of healthy individuals who had been phenotyped for various aspects of memory function. In this study, we have extended the cohort to 597 individuals and examined multiple genetic variants across both the BDNF and BDNF-AS genes for association with visual memory performance as assessed by the Wechsler Memory Scale—Fourth Edition subtests Visual Reproduction I and II (VR I and II). VR I assesses immediate visual memory, whereas VR II assesses long-term visual memory. Genetic association analyses were performed for 34 single nucleotide polymorphisms genotyped on Illumina OmniExpress BeadChip arrays with the immediate and long-term visual memory phenotypes. While none of the BDNF and BDNF-AS variants were shown to be significant for immediate visual memory, we found 10 variants (including the Val66Met polymorphism (p-value = 0.006)) that were nominally associated, and three variants (two variants in BDNF and one variant in the BDNF-AS locus) that were significantly associated with long-term visual memory. Our data therefore suggests a potential role for BDNF, and its anti-sense transcript BDNF-AS, in long-term visual memory performance.

The role of auditory cortices in the retrieval of single-trial auditory-visual object memories

Single-trial encounters with multisensory stimuli affect both memory performance and early-latency brain responses to visual stimuli. Whether and how auditory cortices support memory processes based on single-trial multisensory learning is unknown and may differ qualitatively and quantitatively from comparable processes within visual cortices due to purported differences in memory capacities across the senses. We recorded event-related potentials (ERPs) as healthy adults (n = 18) performed a continuous recognition task in the auditory modality, discriminating initial (new) from repeated (old) sounds of environmental objects. Initial presentations were either unisensory or multisensory; the latter entailed synchronous presentation of a semantically congruent or a meaningless image. Repeated presentations were exclusively auditory, thus differing only according to the context in which the sound was initially encountered. Discrimination abilities (indexed by d') were increased for repeated sounds that were initially encountered with a semantically congruent image versus sounds initially encountered with either a meaningless or no image. Analyses of ERPs within an electrical neuroimaging framework revealed that early stages of auditory processing of repeated sounds were affected by prior single-trial multisensory contexts. These effects followed from significantly reduced activity within a distributed network, including the right superior temporal cortex, suggesting an inverse relationship between brain activity and behavioural outcome on this task. The present findings demonstrate how auditory cortices contribute to long-term effects of multisensory experiences on auditory object discrimination. We propose a new framework for the efficacy of multisensory processes to impact both current multisensory stimulus processing and unisensory discrimination abilities later in time.

Human intracranial high-frequency activity during memory processing: neural oscillations or stochastic volatility?

Intracranial high-frequency activity (HFA), which refers to fast fluctuations in electrophysiological recordings, increases during memory processing. Two views have emerged to explain this effect: (1) HFA reflects a synchronous signal, related to underlying gamma oscillations, that plays a mechanistic role in human memory and (2) HFA reflects an asynchronous signal that is a non-specific marker of brain activation. We review recent data supporting each of these views and conclude that HFA during memory processing is more consistent with an asynchronous signal. Memory-related HFA is therefore best conceptualized as a biomarker of neural activation that can functionally map memory with high spatial and temporal precision.

The Naïve and the Distrustful: state dependency of hippocampal computations in manipulative memory distortion

Flexible mnemonic mechanisms that adjust to different internal mental states can provide a major adaptive advantage. However, little is known regarding how this flexibility is achieved in the human brain. We examined brain activity during retrieval of false memories of a movie, generated by exposing participants to misleading information. Half of the participants suspected the memory manipulation (Distrustful), whereas the other half did not (Naïve). Distrustful displayed more accurate memory performance and a brain signature different than that of Naïve. In Distrustful, the ability to differentiate true from false information was driven by a qualitatively distinct hippocampal activity for endorsed items, consistent with the view that hippocampal encoding allows recollection of a specific source. Conversely, in Naïve, BOLD differences between true and false memories were linearly correlated with accuracy across participants, suggesting that Naïve subjects needed to reinstate and evaluate stored information to discern true from false. We propose that our results lend support to models suggesting that hippocampal activity can exhibit different computational schemes, depending on memorandum attributes. Furthermore, we show that trust, considered as a subjective state of mind, may alter basic hippocampal strategies, influencing the ability to separate real from false memory.

Briefly cuing memories leads to suppression of their neural representations

Previous studies have linked partial memory activation with impaired subsequent memory retrieval (e.g., Detre et al., 2013) but have not provided an account of this phenomenon at the level of memory representations: How does partial activation change the neural pattern subsequently elicited when the memory is cued? To address this question, we conducted a functional magnetic resonance imaging (fMRI) experiment in which participants studied word-scene paired associates. Later, we weakly reactivated some memories by briefly presenting the cue word during a rapid serial visual presentation (RSVP) task; other memories were more strongly reactivated or not reactivated at all. We tested participants' memory for the paired associates before and after RSVP. Cues that were briefly presented during RSVP triggered reduced levels of scene activity on the post-RSVP memory test, relative to the other conditions. We used pattern similarity analysis to assess how representations changed as a function of the RSVP manipulation. For briefly cued pairs, we found that neural patterns elicited by the same cue on the pre- and post-RSVP tests (preA-postA; preB-postB) were less similar than neural patterns elicited by different cues (preA-postB; preB-postA). These similarity reductions were predicted by neural measures of memory activation during RSVP. Through simulation, we show that our pattern similarity results are consistent with a model in which partial memory activation triggers selective weakening of the strongest parts of the memory.

Impairing existing declarative memory in humans by disrupting reconsolidation

During the past decade, a large body of research has shown that memory traces can become labile upon retrieval and must be restabilized. Critically, interrupting this reconsolidation process can abolish a previously stable memory. Although a large number of studies have demonstrated this reconsolidation associated amnesia in nonhuman animals, the evidence for its occurrence in humans is far less compelling, especially with regard to declarative memory. In fact, reactivating a declarative memory often makes it more robust and less susceptible to subsequent disruptions. Here we show that existing declarative memories can be selectively impaired by using a noninvasive retrieval-relearning technique. In six experiments, we show that this reconsolidation-associated amnesia can be achieved 48 h after formation of the original memory, but only if relearning occurred soon after retrieval. Furthermore, the amnesic effect persists for at least 24 h, cannot be attributed solely to source confusion and is attainable only when relearning targets specific existing memories for impairment. These results demonstrate that human declarative memory can be selectively rewritten during reconsolidation.

Learning and Remembering with Others: The Key Role of Retrieval in Shaping Group Recall and Collective Memory

People frequently collaborate to learn and remember information, and this may help groups create a shared representation of the world (i.e., collective memories). However, contrary to intuitions, collaboration also lowers group recall levels. Such impairment occurs regardless of whether people collaborate when first experiencing, or encoding, an event (the collaborative encoding deficit), or when retrieving, or remembering, the event (the collaborative inhibition effect). In understanding how collaboration impairs group recall and enhances collective or shared memories it remains unknown as to where collaboration exerts the greatest influence - at encoding or at retrieval - to shape these distinct phenomena. The current study simultaneously compared collaboration at these two stages and revealed the power of collaborative retrieval. Collaboration impaired the group recall product at both time points, but especially so at retrieval. Furthermore, only collaborative retrieval played a significant role in the formation of collective memories.

BDNF val66met affects hippocampal volume and emotion-related hippocampal memory activity

The val(66)met polymorphism on the BDNF gene has been reported to explain individual differences in hippocampal volume and memory-related activity. These findings, however, have not been replicated consistently and no studies to date controlled for the potentially confounding impact of early life stress, such as childhood abuse, and psychiatric status. Using structural and functional MRI, we therefore investigated in 126 depressed and/or anxious patients and 31 healthy control subjects the effects of val(66)met on hippocampal volume and encoding activity of neutral, positive and negative words, while taking into account childhood abuse and psychiatric status. Our results show slightly lower hippocampal volumes in carriers of a met allele (n=54) relative to val/val homozygotes (n=103) (P=0.02, effect size (Cohen's d)=0.37), which appeared to be independent of childhood abuse and psychiatric status. For hippocampal encoding activity, we found a val(66)met-word valence interaction (P=0.02) such that carriers of a met allele showed increased levels of activation in response to negative words relative to activation in the neutral word condition and relative to val/val homozygotes. This, however, was only evident in the absence of childhood abuse, as abused val/val homozygotes showed hippocampal encoding activity for negative words that was comparable to that of carriers of a met allele. Neither psychiatric status nor memory accuracy did account for these associations. In conclusion, BDNF val(66)met has a significant impact on hippocampal volume independently of childhood abuse and psychiatric status. Furthermore, early adverse experiences such as childhood abuse account for individual differences in hippocampal encoding activity of negative stimuli but this effect manifests differently as a function of val(66)met.

Influences of Source - Item Contingency and Schematic Knowledge on Source Monitoring: Tests of the Probability-Matching Account

The authors investigated conditions under which judgments in source-monitoring tasks are influenced by prior schematic knowledge. According to a probability-matching account of source guessing (Spaniol & Bayen, 2002), when people do not remember the source of information, they match source guessing probabilities to the perceived contingency between sources and item types. When they do not have a representation of a contingency, they base their guesses on prior schematic knowledge. The authors provide support for this account in two experiments with sources presenting information that was expected for one source and somewhat unexpected for another. Schema-relevant information about the sources was provided at the time of encoding. When contingency perception was impeded by dividing attention, participants showed schema-based guessing (Experiment 1). Manipulating source - item contingency also affected guessing (Experiment 2). When this contingency was schema-inconsistent, it superseded schema-based expectations and led to schema-inconsistent guessing.

Brain-derived neurotrophic factor val66met polymorphism affects human memory-related hippocampal activity and predicts memory performance

BDNF plays a critical role in activity-dependent neuroplasticity underlying learning and memory in the hippocampus. A frequent single nucleotide polymorphism in the targeting region of the human BDNF gene (val66met) has been associated with abnormal intracellular trafficking and regulated secretion of BDNF in cultured hippocampal neurons transfected with the met allele. In addition, the met allele has been associated with abnormal hippocampal neuronal function as well as impaired episodic memory in human subjects, but a direct effect of BDNF alleles on hippocampal processing of memory has not been demonstrated. We studied the relationship of the BDNF val66met genotype and hippocampal activity during episodic memory processing using blood oxygenation level-dependent functional magnetic resonance imaging and a declarative memory task in healthy individuals. Met carriers exhibited relatively diminished hippocampal engagement in comparison with val homozygotes during both encoding and retrieval processes. Remarkably, the interaction between the BDNF val66met genotype and the hippocampal response during encoding accounted for 25% of the total variation in recognition memory performance. These data implicate a specific genetic mechanism for substantial normal variation in human declarative memory and suggest that the basic effects of BDNF signaling on hippocampal function in experimental animals are important in humans.

Prefrontal-temporal circuitry for episodic encoding and subsequent memory

Humans encounter and form memories for multiple types of experiences that differ in content, novelty, and memorability. Critical for understanding memory is determining (1) how the brain supports the encoding of events with differing content and (2) whether neural regions that are sensitive to novelty also influence whether stimuli will be subsequently remembered. This event-related functional magnetic resonance imaging (fMRI) study crossed content (picture/word), novelty (novel/repeated), and subsequent memory (remembered/forgotten) to examine prefrontal and temporal lobe contributions to encoding. Results revealed three patterns of encoding-related activation in anatomically connected inferior prefrontal and lateral temporal structures that appeared to vary depending on whether visuospatial/visuo-object, phonological/lexical, or semantic attributes were processed. Event content also modulated medial temporal lobe activity; word encoding predominantly activated the left hemisphere, whereas picture encoding activated both hemispheres. Critically, in prefrontal and temporal regions that were modulated by novelty, the magnitude of encoding activation also predicted whether an event would be subsequently remembered. These results suggest that (1) regions that demonstrate a sensitivity to novelty may actively support encoding processes that impact subsequent explicit memory and (2) multiple content-dependent prefrontal-temporal circuits support event encoding. The similarities between prefrontal and lateral temporal encoding responses raise the possibility that prefrontal modulation of posterior cortical representations is central to encoding.

Selection of currently relevant memories by the human posterior medial orbitofrontal cortex

We have demonstrated previously that patients producing spontaneous confabulations fail to suppress currently irrelevant memory traces, so that they act and think on the basis of a false, temporally displaced (past) reality. All spontaneous confabulators had anterior limbic damage, in particular of the orbitofrontal cortex and basal forebrain. These findings indicated that these structures are essential for distinguishing between mental representations of ongoing reality and currently irrelevant memories. In the present study, we used a similar experimental paradigm as in our clinical studies and H(2)(15)O positron emission tomography to explore the selection of currently relevant memories by the healthy human brain. Subjects were repeatedly presented with the same set of pictures, arranged in different order each time, and were requested to indicate picture recurrences within the runs. Thus, performance in the first run depended on new learning, whereas subsequent runs required the distinction between picture repetitions within the current run ("now") and previous picture presentations in earlier runs. Whereas initial learning activated medial temporal structures, subsequent runs provoked circumscribed posterior medial orbitofrontal activation. We suggest that this area is essential for sorting out mental associations that pertain to ongoing reality.

Age-related differences in memory for lateral orientation of pictures

Two experiments examined memory for the lateral orientation of scenic pictures by young and elderly adults. In Experiment 1, an input list of pictures was followed by a test demanding discrimination between (a) targets versus reversed copies of input items, or (b) targets versus new pictures which verbally resembled input items. The age-related difference was reliably larger in the former task than in the latter. Experiment 2 compared incidental versus intentional acquisition of orientation under conditions of short (1 second) and long (5 second) presentation of pictures at input. With short presentation, though not with long presentation, intentional instructions reliably impaired orientation memory. With both presentation times, robust age-related differences were obtained. The results suggest an age-related deficit in truly non-intentional encoding of orientation, and pose a challenge for capacity theories of memory across the lifespan.