United States

Deep reinforcement learning (Deep RL) methods can train artificial agents (AAs) to reach or exceed human-level performance. However, in multiagent contexts requiring competitive behavior or where the aim is to use AAs for human training, the qualitative behaviors AAs adopt may be just as important as their performance to ensure representative training. This paper compares human behaviors and performance when competing against either a human expert or an AA opponent trained using Deep RL on a 2-dimensional version of Pong. Results show that participants were not sensitive to the movement differences between the human expert and AA. Further, the participants did not alter their behaviors, except to compensate for differences in the environmental states caused by the opponents. The paper concludes with discussion on the potential impacts of AA training on human behavior with regard to representative design in the areas of skill development and team training.

CogSci 2021

Perceptual Sensitivity to an Artificial Co-Actor in Competitive 2D Pong

### Looking for Cogsci 2023? Please [click here](https://underline.io/events/405/reception) to switch to 2023 edition.


**Dear Colleagues,**

Welcome to the 43rd Annual Conference of the Cognitive Science Society! 

This year’s conference brings together the latest research from all Cognitive Science and also highlights this year’s theme **Comparative Cognition: Animal Minds.** 

The program features four plenary speakers: Friederike Range (University of Veterinary Medicine, Vienna), Onur Güntürkün (Ruhr University Bochum), Julia Fischer (University of Göttingen), and Erich Jarvis (The Rockefeller University). The keynotes and invited symposia showcase the broad spectrum of ideas that encompass the conference theme of comparative cognition. 

Further, the program includes the Rumelhart Prize keynote presentation by Susan Goldin-Meadow (University of Chicago), and symposia to celebrate the winner of the Elman Prize: Wei Ji Ma (New York University), the Heineken Award winner Robert Zatorre (McGill University), and the Robert J. Glushko Dissertation Prizes session.

The program committee for CogSci 2021 received 1,025 submissions, including 710 full papers, 275 member abstracts, 8 publication-based talks, as well as 15 proposals for symposia, and 17 for workshops and tutorials. After a rigorous review process, the committee selected 152 papers for oral presentation and inclusion in the conference proceedings (14.8%), 332 papers for poster presentation and inclusion in the proceedings (32.4%), and 175 papers for poster presentation with inclusion of abstracts in the proceedings (17.1%) We also selected 275 submitted member abstracts. In addition, we accepted 6 publication-based talks, 12 symposia, and 9 workshops and tutorials to make for an exciting and inclusive program.

We hope that you enjoy the conference this year!<br><br>

**Your Program Chairs,**
![](https://assets.underline.io/uploads/markdown_image/1/image/99984908caba5e6453e5fc5ba34f3507.jpg)

**Learn more about the Vienna Cognitive Science Hub at the University of Vienna, hosts of CogSci 2021**

<iframe width="560" height="315" src="https://www.youtube.com/embed/464h1m3iU68" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>

To review the proceedings for the conference or review the searchable program, please visit the [website](https://cognitivesciencesociety.org/program/).


The Cognitive Science Society brings together researchers from around the world who hold a common goal: understanding the nature of the human mind. The mission of the Society is to promote Cognitive Science as a discipline, and to foster scientific interchange among researchers in various areas of study, including Artificial Intelligence, Linguistics, Anthropology, Psychology, Neuroscience, Philosophy, and Education. The Society is a non-profit professional organization and its activities include sponsoring an annual conference and publishing the journals Cognitive Science and TopiCS.

poster

Visual perception initiated with a low-level derivation of Spatio-temporal edges and advances to a higher-level perception of filled surfaces. According to the isomorphic theory, this perceptual filling-in is governed by an activation spread across the retinotopic map, driven from edges to interiors. Here we propose two biologically plausible spiking neural networks, which demonstrate perceptual filling-in by resolving the Poisson equation. Each network exhibits a distinct dynamic and architecture and could be realized and further integrated in the brain.

Biologically Plausible Spiking Neural Networks for Perceptual Filling-In

This preregistered study examined whether extraneous illustration details promote attentional competition and hinder reading comprehension in beginning readers. Reading comprehension was highest in the Streamlined Condition (text + relevant illustrations) compared to a Standard Condition (text + relevant illustrations + extraneous illustrations) and Text Only Condition (no illustrations). Gaze shifts away from the text were highest in the Standard Condition, indicating increased distractibility while reading text with extraneous illustration details. Gaze shifts away from the text were associated with performance on an independent measure of attention, validating eye gaze patterns as an assessment of attentional allocation while reading. Lower comprehension in the Standard Condition was associated with higher gaze shifts away from text and lower scores on the independent measure of selective attention. This study suggests that illustrations can support reading comprehension, but only when they are optimally designed. Importantly, the removal of extraneous details did not decrease book enjoyment.

The Optimal Amount of Visuals Promotes Children’s Comprehension and Attention: An Eye Tracking Study

We present the first analysis of multi-frequency neural entrainment to dynamic visual features which drives sign language comprehension. Using the measure of EEG coherence to optical flow in video stimuli, we are able to classify fluent signers’ brain states as denoting online language comprehension, or non-comprehension during watching of non-linguistic videos that are equivalent in low-level spatiotemporal features and high-level scene parameters. The data also indicates that lower frequencies, such as 1 Hz and 4 Hz, contribute substantially to brain state classification, indicating relevance of neural coherence to the signal at these frequencies to language comprehension. These findings suggest that fluent signers rely on predictive processing during online comprehension.

Application of machine learning to signal entrainment identifies predictive processing in sign language.

The present paper addresses the study of cross-linguistic phonosemantic correspondences within a deep learning framework. An LSTM-based Recurrent Neural Network is trained to associate the phonetic representation of a word, encoded as a sequence of feature vectors, to its corresponding semantic representation in a multilingual and cross-family vector space. The processing network is then tested, without further training, in a language that does not appear in the training set and belongs to a different language family. The performance of the model is evaluated through a comparison with a monolingual and mono-family upper bound and a randomized baseline. After the assessment of the network's performance, the distribution of phonosemantic properties in the lexicon is inspected in relation to different (psycho)linguistic variables, showing a link between lexical non-arbitrariness and semantic, syntactic, pragmatic, and developmental factors.

A Layered Bridge from Sound to Meaning: Investigating Cross-linguistic Phonosemantic Correspondences

From a young age, we can select actions to achieve desired goals, infer the goals of other agents, and learn causal relations in our environment through social interactions. Crucially, these abilities are productive and generative: we can impute desires to others that we have never held ourselves. These abilities are often captured by only partially overlapping models, each requiring substantial changes to fit combinations of abilities. Here, in an attempt to unify previous models, we present a neural network underpinned by the linearly solvable Markov Decision Process (LMDP) framework which permits a distributed representation of tasks. The network contains two pathways: one captures the desirability of states, and another encodes the passive dynamics of state transitions in the absence of control. Interactions between pathways are bound by a principle of rational action, enabling generative inference of actions, goals, and causal relations supported by gradient updates to parts of the network.

Inferring Actions, Intentions, and Causal Relations in a Deep Neural Network

The question of how algorithms in general and cognitive skills in particular are implemented by our nervous system is at the core of cognitive science. The notions of what it means for a physical system (such as our nervous system) to implement an algorithm, however, are surprisingly vague. We argue that a rigorous theory is needed to formulate and evaluate precise hypotheses about the brain's cognitive functions and propose a definition of the term algorithm as a chain of functions. Subsequently, we define the term implementation via a sequence of projections from a dynamical system, represented by a Markov process, to the algorithm. We furthermore show the practical applicability of this approach in a simulated example. We believe that the theory proposed here contributes to bridging the gap between the algorithmic and the implementational level by rendering the task at hand theoretically precise.

A theory of algorithms and implementations and their relevance to cognitive science

The N400 component of the event related brain potential is widely used to investigate language and meaning processing. However, despite much research the component’s functional basis remains actively debated. Recent work showed that the update of the predictive representation of sentence meaning (semantic update, or SU) generated by the Sentence Gestalt model (McClelland, St. John, & Taraban, 1989) consistently displayed a similar pattern to the N400 amplitude in a series of conditions known to modulate this event-related potential. These results led (Rabovsky, Hansen, and McClelland,2018) to suggest that the N400 might reflect change in a probabilistic representation of meaning corresponding to an implicit semantic prediction error. However, a limitation of this work is that the model was trained on a small artificial training corpus and thus could not be presented with the same naturalistic stimuli presented in empirical experiments. In the present study, we overcome this limitation and directly model the amplitude of the N400 elicited during naturalistic sentence processing by using as predictor the SU generated by a Sentence Gestalt model trained on a large corpus of texts. The results reported in this paper corroborate the hypothesis that the N400 component reflects the change in a probabilistic representation of meaning after every word presentation. Further analyses demonstrate that the SU of the Sentence Gestalt model and the amplitude of the N400 are influenced similarly by the stochastic and positional properties of the linguistic input.

Predicting the N400 ERP component using the Sentence Gestalt model trained on a large scale corpus

We propose a conceptual framework of multiversional narrative processing. Multiversional narrative processing is the consideration of multiple possible event sequences for an incomplete narrative during reception. It occurs naturally and is experienced in a wide range of cases, such as suspense, surprise, counterfactuals, and detective stories. Receiving a narrative, we propose, is characterized by the spontaneous creation of competing interpretive versions of the narrative that are then used to create predictions and projections for the narrative’s future. These predictions serve as a mechanism for integrating incoming information and updating the narrative model through prediction error, without completely eliminating past versions. We define this process as having three aspects: (1) constrained expectations, (2) preference projection, and (3) causal extrapolation. Constrained expectations and preference projections respectively create the bounds and subjective desires for a narrative’s progress, while causal extrapolation builds, reworks, and maintains the potential models for understanding the narrative.

Multiversionality: Considering multiple possibilities in the processing of narrative

The N400 brain potential has been used as a neural correlate of meaning-related processing in the brain, but its underlying computational mechanism is still not well understood. Although efforts to model the N400 as an update of a probabilistic representation of meaning have been promising, the limited scope of earlier models has restricted experiments to highly simplified sentences. Here, we expand modelling of the N400 to naturalistic sentences using a large-scale, state-of-the-art deep learning language model. We investigate the correspondence between updates in the internal state of the model and the N400 in one quantitative experiment and four qualitative experiments. Our findings suggest that activation updates in the model correspond to several N400 effects, but cannot account for all of them.

Meaning in brains and machines: Internal activation update in large-scale language model partially reflects the N400 brain potential

Recent advances in neural architectures have revived the problem of morphological rule learning. We evaluate the Transformer as a model of morphological rule learning and compare it with Recurrent Neural Networks (RNN) on English, German, and Russian. We bring to the fore a hitherto overlooked problem, the morphological gaps, where the expected inflection of a word is missing. For example, 63 Russian verbs lack a first-person-singular present form such that one cannot comfortably say "*oščušču" ("I feel"). Even English has gaps, such as the past participle of "stride": the function of morphological inflection can be partial. Both neural architectures produce inflections that ought to be missing. Analyses reveal that Transformers recapitulate the statistical distribution of inflections in the training data, similar to RNNs. Models' success on English and German is driven by the fact that rules in these languages can be identified with the majority forms, which is not universal.

Perceptual Sensitivity to an Artificial Co-Actor in Competitive 2D Pong

Downloads

Next from CogSci 2021

Biologically Plausible Spiking Neural Networks for Perceptual Filling-In

Similar lecture

Learning from Agentic Actions: Modelling Causal Inference from Intention