Cognitive Psychology has traditionally examined how information is represented, and the processes that operate on those representations. Contemporary computational models of spoken word recognition adequately capture the processes involved in recognizing words, but fail to account for the influence on processing of the large-scale structure that exists among lexical representations. We compare the ability of a contemporary model of spoken word recognition (jTRACE; Strauss et al, 2007) to the ability of a network model with a spreading activation-like process (spreadr; Siew, 2019) to account for the findings from several previously published behavioral studies of language processing. The results suggest that viable models of spoken word recognition must account for the influence on processing of the large-scale structure that exists among lexical representations.

The mental lexicon can be represented as a language network of words that are connected if they are phonologically and/or orthographically similar to each other. Here we investigate whether structure of the phonographic network, where orthographically and phonologically similar words are connected to each other, affected visual word recognition in languages less frequently examined within psycholinguistics. Network measures that quantify the orthographic, phonological, and phonographic similarity structure of words were calculated for English, Dutch, French, and Spanish. Using behavioral data from visual word recognition megastudies, regression analyses examined the influence of orthographic, phonological, and phonographic similarity on reaction times and accuracies in visual lexical decision. Across all languages, phonographic similarity structure accounted for additional variance in lexical decision performance, even after accounting for the independent contributions of orthographic and phonological similarity. Specifically, phonographic similarity facilitated lexical decision performance in English, Dutch and Spanish, but inhibited lexical decision performance in French.

Scientists are limited by their assumptions, and more so by those they fail to recognize. In this talk, I will highlight several limiting assumptions I see in cognitive network science by way of speculating about possible alternatives. These include maps, vehicles, wormholes, particle-wave problems, and skyhooks, which each offer challenges and potential solutions to garden variety network science.

We do not simply have concepts; we use concepts. And, the way in which we use concepts can dynamically change the relations among them. One way to shed light on this dynamic nature is to examine how the novel processing of concepts—in our case, interpreting unfamiliar nominal combinations—might reconfigure semantic memory networks. To achieve this, a computational network science methodology was applied to examine how the act of interpreting novel conceptual combinations affects properties of semantic networks (e.g., connectivity). Specifically, we examine two processes by which concepts are combined,either by using an attribute of one concept to describe another (attributive) or by forming some relation between two concepts (relational).Using a free association task, participants’ semantic networks were estimated and compared before and after either a relational combination, attributive combination, or a baseline condition (that did not include the conceptual combination task). We find that relational, but not attributive, combinations increased the connectivity and lowered structure in the network. This effect may indicate that interpretation of relational combinations requires the generation of novel contexts, thus leading to greater re-structuring of the semantic network.

This satellite should ideally fit within half a day, i.e. 4 hours maximum. In order to keep reasonable attention spans and include as many contributed talks as possible within the above time limit, both invited talks and contributed talks will be shorter than in mainstream in-person satellites.

Contributed talks will last 10 mins + 5 mins discussion (Q&A) each, whereas invited talks will last for 15 mins + 5 mins Q&A each. Each member of the Scientific Committee is invited also to provide, optionally, an invited talk, so that the total timing devoted to invited talks can be a maximum of 5 * 20 min = 1h 40 mins.

The call for contributed talks is currently CLOSED. Relevant topics include, but are not limited to:

• Complex systems and cognition;

• Social media inquiry under the lens of cognitive science;

• Artificial intelligence, machine learning and cognitive modelling;

• Network models of the human brain and/or mind;

• Big Data, personality traits and human dynamics;

• System thinking, clinical populations and mental health

Satellite registration and potential contributions are managed through a Google Form (see bottom page). The Committee will sift abstracts. Each Committee member will vote for acceptance or rejection of each potential contribution. Only contributions unanimously accepted by the Committee will be accepted for presentation. The Main Organiser will not vote for acceptance but will gather and coordinate votes and then notify acceptance/rejection. Authors of rejected submissions might still produce a link to a poster or an online paper that will be hosted on the satellite website (but not presented during the event).