Sources of variability in linguistic memory systems

In sentence comprehension, what are the cognitive constraints that determine number agreement computation? Two broad classes of theoretical proposals are: (i) Representation distortion accounts, which assume that the number feature on the subject noun gets overwritten probabilistically by the number feature on a non-subject noun, leading to a non-veridical memory trace of the subject noun; and (ii) The cue-based retrieval account, a general account of dependency completion processes which assumes that the features on the subject noun remain intact, and that processing difficulty is only a function of the memory constraints on dependency completion. However, both these classes of model fail to account for the full spectrum of number agreement patterns observed in published studies. Using 17 benchmark datasets on number agreement from four languages, we implement seven computational models: three variants of representation distortion, two cue-based retrieval models, and two hybrid models that assume both representation-distortion and retrieval. Quantitative model comparison shows that the best fit is achieved by a hybrid model that assumes both feature distortion (specifically, feature percolation) and cue-based retrieval; numerically, the second-best quantitative fit was achieved by a distortion-based model of number attraction that assumes grammaticality bias during reading. More broadly, the work furnishes comprehensive evidence to support the idea that cue-based retrieval theory, which aims to be a general account of dependency completion, needs to incorporate a feature distortion process.

The Functional Impact of Mental Imagery on Conscious Perception

Low-power content addressable memory (CAM) array for mobile devices

Alpha power decreases associated with prediction in written and spoken sentence comprehension

Understanding language requires applying cognitive operations (e.g., memory retrieval, prediction, structure building) that are relevant across many cognitive domains to specialized knowledge structures (e.g., a particular language’s lexicon and syntax). Are these computations carried out by domain-general circuits or by circuits that store domain-specific representations? Recent work has characterized the roles in language comprehension of the language network, which is selective for high-level language processing, and the multiple-demand (MD) network, which has been implicated in executive functions and linked to fluid intelligence and thus is a prime candidate for implementing computations that support information processing across domains. The language network responds robustly to diverse aspects of comprehension, but the MD network shows no sensitivity to linguistic variables. We therefore argue that the MD network does not play a core role in language comprehension and that past findings suggesting the contrary are likely due to methodological artifacts. Although future studies may reveal some aspects of language comprehension that require the MD network, evidence to date suggests that those will not be related to core linguistic processes such as lexical access or composition. The finding that the circuits that store linguistic knowledge carry out computations on those representations aligns with general arguments against the separation of memory and computation in the mind and brain.

III.4 - Hopfield Model and Optimization Problems

Frontiers Events is a rapidly growing calendar management system dedicated to the scheduling of academic events. This includes announcements and invitations, participant listings and search functionality, abstract handling and publication, related events and post-event exchanges. Whether an organizer or participant, make your event a Frontiers Event!

This study investigated the nature of the underlying working memory system supporting sentence processing through examining individual differences in sensitivity to retrieval interference effects during sentence comprehension. Interference effects occur when readers incorrectly retrieve sentence constituents which are similar to those required during integrative processes. We examined interference arising from a partial match between distracting constituents and syntactic and semantic cues, and related these interference effects to performance on working memory, short-term memory (STM), vocabulary, and executive function tasks. For online sentence comprehension, as measured by self-paced reading, the magnitude of individuals' syntactic interference effects was predicted by general WM capacity and the relation remained significant when partialling out vocabulary, indicating that the effects were not due to verbal knowledge. For offline sentence comprehension, as measured by responses to comprehension questions, both general WM capacity and vocabulary knowledge interacted with semantic interference for comprehension accuracy, suggesting that both general WM capacity and the quality of semantic representations played a role in determining how well interference was resolved offline. For comprehension question reaction times, a measure of semantic STM capacity interacted with semantic but not syntactic interference. However, a measure of phonological capacity (digit span) and a general measure of resistance to response interference (Stroop effect) did not predict individuals' interference resolution abilities in either online or offline sentence comprehension. The results are discussed in relation to the multiple capacities account of working memory (e.g., Martin and Romani, 1994; Martin and He, 2004), and the cue-based retrieval parsing approach (e.g., Lewis et al., 2006; Van Dyke et al., 2014). While neither approach was fully supported, a possible means of reconciling the two approaches and directions for future research are proposed.

Review of: Watrous AJ, Tandon N, Conner CR, Pieters T, Ekstrom AD. Frequency-specific network connectivity increases underlie accurate spatiotemporal memory retrieval. Nat Neurosci 2013 doi:10.1038/nn.3315. Humans depend on a memory system to rapidly and veraciously store the massive stream of incoming information documenting their experiences, and faithfully represent and recall that information for future use. Devastating dysfunction in the system is observed in schizophrenia and Alzheimer’s disease. While the medial temporal lobe (MTL), prefrontal cortex (PFC), and parietal lobe have been implicated in an episodic memory retrieval network, the processes underlying successful retrieval remain undetermined, particularly regarding simultaneous recall of different classes of information represented by shared neuronal ensembles. Leading hypotheses propose that retrieval may be defined by changes in connectivity and activity of one region in the network, connections throughout the network, or phase-synchronized activity at specific frequencies. In a recent study, Watrous et al. tested these theories with simultaneous electrocortigraphical (ECoG) recordings from these regions during episodic memory retrieval. Six patients with medically intractable epilepsy underwent invasive monitoring, and the derived data were used for this study. Electrodes were localized to subregions of the MTL, PFC, and parietal cortex. Patients performed a spatiotemporal memory-dependent task with a virtual taxi game, in which they drove passengers to five irregularly spaced stores in a specific sequential order. Once patients reached criteria levels of accuracy or training time, retrieval of spatial and temporal information was tested separately. Phase synchronization was utilized as a measure of functional connectivity to test the different hypotheses. Pairwise phase consistency (PPC) index, which is based on phase angle difference between signals from different electrodes, was used to quantify the phase synchronization across electrode pairs. Two subregions, termed nodes within the network, were considered functionally connected when the PPC between them differed significantly depending on the condition (correct vs. incorrect responses, spatial vs. temporal memory retrieval). The node degree, a node’s total number of connections, also marked the magnitude of connectivity. Nodes with the highest node degree were defined as hubs. With the greatest connectivity, a hub is thought to play a central role in mediating interactions across the network for retrieval. The percent connectivity was also calculated as the number of connections a node has in a given condition out of the total possible connections. The authors found that functional connectivity, derived by measuring phase-synchronized low-frequency oscillations, was augmented both globally and in the parahippocampal gyrus (PHG) with successful retrieval of spatiotemporal memory. PPC of delta-theta frequency (1-10 Hz) range was greater for correct responses than incorrect. While percent connectivity increased throughout the network with successful retrieval, a disproportionately greater increase in the PHG node was also observed. PHG was designated a hub in the retrieval network. According to these findings, both network-wide and regional, but not frequency-specific, increases in functional connectivity occur for retrieval of spatiotemporal information. Comparing spatial versus temporal memory retrieval uncovered spectral, temporal, and regional segregation of functional connectivity within the retrieval network. For retrieval of spatial information, PPC and percent connectivity were increased between PHG, superior frontal gyrus (SFG), and middle frontal gyrus (MFG), and precuneus areas, at 1-4 Hz, and early in memory retrieval. Temporal memory retrieval was characterized by greater PPC and percent connectivity between PHG, SFG, MFG, and the inferior parietal lobule, at 7-10 Hz, and later in memory retrieval. The PHG was a hub in both conditions, making the same number of overall connections with other nodes, and with differential clustering of connectivity (Figure 1).Figure 1: Frequency-specific synchronization occurs among several components of the human memory system during context retrieval. (G-J) Correct spatial and temporal retrieval networks (within the context of the virtual taxi game task) in the 1–4-Hz and 7–10-Hz bands.This work has notable implications for neuroscience and neurosurgery. The role of mesial temporal lobe structures (particularly the hippocampus) in memory recall remains controversial, particularly after memory consolidation. The hippocampus’s role in memory encoding is well established. Theories of consolidation typically argue that retrieval of consolidated information is independent of the hippocampus, while the Multiple Trace Theory (MTT) proposes that memory recall is hippocampus-dependent. The identification of the PHG, which is intimately connected with the hippocampus, as a hub, lends indirect support for MTT. Since the PHG includes connections from entorhinal, parahippocampal, and perirhinal cortices, but not the hippocampus proper, the evidence does not directly indicate specific action of the hippocampus but rather the MTL in general. Additionally, processes supporting functional connectivity and recall of distinct types of information were elucidated by the authors. This discovery could lead to clinical interventions targeting specific types of cognitive processes including memory retrieval. As neurosurgery continues to expand toward treating cognitive disorders, a network-level understanding of mechanisms and components of memory systems will prove invaluable.

Models of human sentence processing effort tend to focus on costs associated with retrieving structures and discourse referents from memory (memory-based) and/or on costs associated with anticipating upcoming words and structures based on contextual cues (expectation-based) (Levy,2008). Although evidence suggests that expectation and memory may play separable roles in language comprehension (Levy et al., 2013), theories of coreference processing have largely focused on memory: how comprehenders identify likely referents of linguistic expressions. In this study, we hypothesize that coreference tracking also informs human expectations about upcoming words, and we test this hypothesis by evaluating the degree to which incremental surprisal measures generated by a novel coreference-aware semantic parser explain human response times in a naturalistic self-paced reading experiment. Results indicate (1) that coreference information indeed guides human expectations and (2) that coreference effects on memory retrieval may exist independently of coreference effects on expectations. Together, these findings suggest that the language processing system exploits coreference information both to retrieve referents from memory and to anticipate upcoming material.

Stress-induced bias of multiple memory systems during retrieval depends on training intensity

Language processing requires us to integrate incoming linguistic representations with representations of past input, often across intervening words and phrases. This computational situation has been argued to require retrieval of the appropriate representations from memory via a set of features or representations serving as retrieval cues. However, even within in a cue-based retrieval account of language comprehension, both the structure of retrieval cues and the particular computation that underlies direct-access retrieval are still underspecified. Evidence from two event-related brain potential (ERP) experiments that show cue-based interference from different types of linguistic representations during ellipsis comprehension are consistent with an architecture wherein different cue types are integrated, and where the interaction of cue with the recent contents of memory determines processing outcome, including expression of the interference effect in ERP componentry. I conclude that retrieval likely includes a computation where cues are integrated with the contents of memory via a linear weighting scheme, and I propose vector addition as a candidate formalization of this computation. I attempt to account for these effects and other related phenomena within a broader cue-based framework of language processing.

The question of longitudinal hippocampal functional specialization is critical to human episodic memory because an accurate understanding of this phenomenon would impact theories of mnemonic function and entail practical consequences for the clinical management of patients undergoing temporal lobe surgery. The implementation of the robotically assisted stereo electroencephalography technique for seizure mapping has provided our group with the opportunity to obtain recordings simultaneously from the anterior and posterior human hippocampus, allowing us to create an unparalleled data set of human subjects with simultaneous anterior and posterior hippocampal recordings along with several cortical regions. Using these data, we address several key questions governing functional hippocampal connectivity in human memory. First, we ask whether functional networks during episodic memory encoding and retrieval are significantly different for the anterior versus posterior hippocampus (PH). We also examine how connections differ across the 2–5 Hz versus 4–9 Hz theta frequency ranges, directly addressing the relative contribution of each of these separate bands in hippocampal–cortical interactions. While we report some overlapping connections, we observe evidence of distinct anterior versus posterior hippocampal networks during memory encoding related to frontal and parietal connectivity as well as hemispheric differences in aggregate connectivity. We frame these findings in light of the proposed AT/PM memory systems. We also observe distinct encoding versus retrieval connectivity patterns between anterior and posterior hippocampal networks, we find that overall connectivity is greater for the PH in the right hemisphere, and further that these networks significantly differ in terms of frontal and parietal connectivity. We place these findings in the context of existing theoretical treatments of human memory systems, especially the proposed AT/PM system. During memory retrieval, we observe significant differences between slow-theta (2–5 Hz) and fast-theta (4–9 Hz) connectivity between the cortex and hippocampus. Taken together, our findings describe mnemonically relevant functional connectivity differences along the longitudinal axis of the human hippocampus that will inform interpretation of models of hippocampal function that seek to integrate rodent and human data.

Drugs and human memory (part 2). Clinical, theoretical, and methodologic issues.

In psycholinguistic research, models of sentence processing prevail which explain human parsing preferences with a set of universal principles derived either from general features of the architecture of the human language processor (e.g., Frazier & Fodor, 1978; Konieczny, 1996), or universal features of grammar (e.g., Pritchett, 1992; Gorrell, 1995), or combinations of both (Gibson, 1991). These accounts usually see the human language processor as a modular subsystem (or a set of subsystems, e.g., Frazier, 1990a) of the cognitive machinery (Fodor, 1983). As Pickering, Crocker, & Clifton (this volume) point out, psycholinguists try to uncover the basic architectures and mechanisms of this module, or more pointedly, the natural laws of the human language processor. These laws, i.e., the basic principles of language processing, should be valid across individuals as well as across languages. The simplest interpretation of this approach to sentence processing implies that similar constructions should be parsed in similar ways irrespective of interindividual or interlingual differences.