Search

Intentional visual imagery is a component of numerous aspects of cognition. Related to visual imagery, mental simulation plays a role in embodied theories of language comprehension that propose activation of modality-specific regions of the brain takes place as part of people understanding language. The extent to which the processes underlying conscious, voluntary visual imagery versus less conscious, more automatic mental simulation overlap is unclear. We investigated this issue by having aphantasics (people who are unable to experience conscious voluntary visual imagery) and control participants perform a property verification task in which they were asked whether a property is a physical part of an object (e.g., lion-tail). We manipulated the false trials in that the two words either were associated (semantically related) but did not form an object-part combination (monkey-banana), or were not associated (apple-cloud). Solomon and Barsalou (2004) demonstrated that word association influenced responses when the words in the false trials were not associated, whereas when they were related, perceptual measures most strongly influenced the results, indicating mental simulation. Here control participants and aphantasics demonstrated similar evidence of the use of both mental simulation and word association when verifying whether the words formed an object-part combination. These results provide evidence that visual imagery and mental simulation are at least somewhat separable cognitive processes.

To what degree does semantic foraging probe semantic network structure? We use a combination of foraging experiments (animals, concrete nouns) and simulations on networks based on nine approaches to semantic similarity to address this question. In data and simulations, we find a significant bias towards naming semantically similar items, and significant correlations between inter-naming time and semantic distance. In previous foraging experiments, a roughly power law distribution with a Lévy range exponent was found in the distribution of inter-naming intervals. We find the value of this exponent is not universal but is sensitive to the search space size in that the exponent decreases (moving further into the Lévy range) as the number of nameable items is exhausted. Moreover, these exponents are not unique to semantic networks but appear in censored random walks on other graphs. Our combined experimental results and simulations provide insights into the topology of semantic memory.

How the mind represents event knowledge, a persons knowledge of events and situations in the world, is the subject ofcompeting theories. Proposals range from an event being represented as a linear order of activities, to a hierarchical struc-ture of scenes of related activities, or in a more fluid computational framework. Additionally, atypical event knowledgeis thought to correlate with Autism Spectrum Disorder. 140 participants (20 per event) ordered normed activity lists for80 events (e.g., taking money out of an ATM, going to a professional baseball game, baking a cake). Network analysessuggest that the temporal structure of events is rich, not strictly linear, and varies across individuals. Furthermore, wecomputed a consensus ordering for each event from participants activity sequences. We calculated deviations from thatordering for each participant, and correlated deviations with a battery of trait inventories to further investigate differencesamong individuals representations of event structure.

Distributional Semantics Models (DSMs) are a primary method for distilling semantic information from corpora. However,a key question remains: What types of semantic relations do DSMs detect? Prior work has addressed this question using alimited set of ratings that typically are either amorphous (association norms) or restricted to semantic similarity (SimLex,SimVerb). We tested four DSMs (SkipGram, CBOW, GloVe, PPMI) using multiple hyperparameters on a theoretically-motivated, rich set of relations involving words from multiple syntactic classes spanning the abstract-concrete continuum(21 sets of ratings). Results show wide variation in the DSMs’ ability to account for the ratings, and that hyperparametertuning buys comparatively little for improving correlations. For CBOW and SkipGram, we included word and contextembeddings. For SkipGram, there was a marked improvement in simulating the human data by averaging them. Ourresults yield important insights into the types of semantic relations that are captured by DSMs.

Representations formed during language comprehension do not always accurately reflect the linguistic input, but aresometimes just good enough (Ferreira et al., 2003). Here, we examined the electrophysiological correlates of such heuristicprocessing. Participants were presented with passive sentences where the plausibility of the fillers of the agent and patientthematic roles was manipulated. As expected, they made more errors in the interpretation of implausible sentences (e.g.,The doctor was treated by the patient). Intriguingly, N400 amplitudes patterned with (mis)interpretation, with increasedamplitudes to the second noun in correctly processed implausible sentences, and equally small amplitudes in plausiblesentences and in incorrectly interpreted implausible sentences. These results are in line with the view that N400 amplitudesreflect the change in an initial heuristic representation of sentence meaning (Rabovsky et al., 2018), but seem difficult toexplain by accounts suggesting that the N400 reflects lexical retrieval (Brouwer et al., 2017).

There are several competing theories regarding how event knowledge is represented in the mind, ranging from a strictlytemporally ordered list of activities to sets of connected scenes which may themselves consist of ordered activities. Weemployed a network science approach to provide data-driven insight into event structure. We converted sets of humangenerated activity sequences, in which roughly 25 participants list up to 12 activities for 81 different events (making asandwich, cleaning the house, taking money out of an ATM, etc.), into directed, weighted networks. Analyses of the eventnetworks revealed a complex and varied temporal structure to events. In addition, we were able to identify scenes withinevents, and use graph theory to understand activity centrality, popularity, and influence, as well as the coupling betweenthese activity characteristics. In the aggregate, we find that network science makes multiple data-driven, empiricallytestable predictions about event structure.

People predict incoming words during online sentencecomprehension based on their knowledge of real-world eventsthat is cued by preceding linguistic contexts. We used thevisual world paradigm to investigate how event knowledgeactivated by an agent-verb pair is integrated with perceptualinformation about the referent that fits the patient role. Duringthe verb time window participants looked significantly more atthe referents that are expected given the agent-verb pair.Results are consistent with the assumption that event-basedknowledge involves perceptual properties of typicalparticipants. The knowledge activated by the agent iscompositionally integrated with knowledge cued by the verbto drive anticipatory eye movements during sentencecomprehension based on the expectations associated not onlywith the incoming word, but also with the visual features of itsreferent.

We present a connectionist model of event knowledge that istrained on examples of sequences of activities that are notexplicitly labeled as events. The model learns co-occurrencepatterns among the components of activities as they occur inthe moment (entities, actions, and contexts), and also learns topredict sequential patterns of activities. In so doing, the modeldisplays behaviors that in humans have been characterized asexemplifying inferencing of unmentioned event components,the prediction of upcoming components (which may or maynot ever happen or be mentioned), reconstructive memory,and the ability to flexibly accommodate novel variations frompreviously encountered experiences. All of these behaviorsemerge from what the model learns.

In previous work investigating the relationship between domain knowledge (of the fictional world of Harry Potter) and sentence comprehension, domain knowledge had a greater impact on electrical brain potentials to words which completed sentences about fictional "facts" participants reported they did not know compared to facts they did. This suggests that individuals use domain knowledge continuously to activate relevant/related concepts as they process sentences, even with only partial knowledge. As that study relied on subjective reports, it may have resulted in response bias related to an individual's overall domain knowledge. In the present study, we therefore asked participants with varying degrees of domain knowledge to complete sentences describing fictional "facts" as an objective measure of sentence‐level knowledge. We then recorded EEG as the same individuals (re‐)read the same sentences, including their appropriate final words, and sorted these according to their objective knowledge scores. Replicating and extending Troyer et al., domain knowledge immediately facilitated access to meaning for unknown words; greater domain knowledge was associated with reduced N400 amplitudes for unknown words. These findings constitute novel evidence for graded preactivation of conceptual knowledge (e.g., at the level of semantic features and/or relations) in the absence of lexical prediction. Knowledge also influenced post‐N400 memory/integration processes for these same unknown words; greater domain knowledge was associated with enhanced late positive components (LPCs), suggesting that deeper encoding during language processing may be engendered when knowledgeable individuals encounter an apparent gap in their knowledge. We examined how content/domain expertise influences electrophysiology during language processing. Within a third of a second, neural signatures indexing access to word meaning reveal sensitivity to fine‐grained variation in domain knowledge—especially for words not explicitly recalled during a fact‐completion task. For the same words, domain knowledge also influenced neural signatures thought to index elaborative encoding processes, likely because generally knowledgeable individuals could more readily link "unknown" words with existing knowledge structures. [ABSTRACT FROM AUTHOR]

We provide a computational account of the integration of various constraints proposed to be involved in the resolution of the direct object/sentence complement ambiguity. In the first part, competition-integration simulations show that a constraint-based model accounts for the results of Garnsey, Pearlmutter, Myers, and Lotocky (1997) at least as well as the garden-path model. In the second part, we compare the efficacy of norming techniques for capturing plausibility effects. Simulations show that norms designed to tap people's conceptual knowledge of events better capture plausibility effects than do norms that are biased toward tapping linguistic knowledge. We conclude that local information concerning event plausibility is an important constraint for understanding ambiguity resolution.

In an attractor model of semantic memory, semantic similarity is determined by degree of featural overlap. In contrast, in spreading activation theory, two concepts are similar if they share features or if they are linked to the same superordinate category node. We present an attractor network model of computing word meaning and use it to simulate the data of McRae and Boisvert (in press), who found that short SOA semantic similarity priming directly depends on degree of featural overlap. The two accounts of semantic similarity are then contrasted in a human experiment. In support of attractor networks, priming effects were determined by featural overlap, and no evidence was found for priming through a purported superordinate node. It is concluded that lexical concepts are not represented as static nodes in a hierarchical system.

Behavioral experiments have demonstrated that peopleencode knowledge of correlations among semantic prop-erties of entities and that this knowledge influences per-formance on semantic tasks (McRae, 1992; McRae, de Sa,& Seidenberg, 1993). Independently, in connectionist the-ory, it has been claimed that relationships among seman-tic properties may provide structure that is required forthe relatively arbitrary mapping from word form to wordmeaning (Hinton&Shallice, 1991). W e explored these is-sues by implementing a modified Hopfield network (1982,1984) to simulate the computation from word form tomeaning. The model was used as a vehicle for developingexplanations for the role played by correlated properties indetermining short interval semantic priming effects and indetermining the ease with which a property is verified aspart of a concept. Simulations of the priming and propertyverification experiments of McRae (1992) are reported. Itis concluded that correlations among properties encodedin conceptual memory play a key role in the dynamics ofthe computation of word meaning. Furthermore, a modelin which property intercorrelations are central to formingbasins of attraction corresponding to concepts may pro-vide important insights into lexical memory.

Property intercorrelations are viewed as central to the representation and processing of real-world object concepts. In contrast, prior research into real-world object concepts has incorporated the assumption that properties are independent and additive. In two studies, the role of correlated properties was explored. Property norms had been collected for 190 natural kinds and artifacts. In Experiment 1, property intercorrelations influenced performance in a property verification task. In Experiment 2, concept similarity, as measured by overlap of independent properties, predicted short interval priming latency for artifacts. In contrast, concept similarity, as measured by overlap of correlated property pairs, predicted short interval priming for natural kinds. The influence of property intercorrelations was stronger for natural kinds because they tended to contain a higher proportion of correlated properties. It was concluded that people encode knowledge about independent and correlated properties of real-world objects. Presently, a Hopfield netwoiic is being implemented to explore implications of allowing a system to encode property intercorrelations. Finally, results suggest that semantic relatedness can be defined in terms of property overlap between concepts.