United States

Team interaction involves the division of labor and coordination of actions between members to achieve a shared goal. Although the dynamics of interactions that afford effective coordination and performance have been a focus in the cognitive science community, less is known about how to generate these flexible and adaptable coordination patterns. This is important when the goal is to design artificial agents that can augment and enhance team coordination as synthetic teammates. Although previous research has demonstrated the negative impact of model-based agents on the pattern of interactions between members using recurrence quantification methods, more recent work utilizing deep reinforcement learning has demonstrated a promising approach to bootstrap the design of agents to team with humans effectively. This paper explores the impact of artificial agent design on the interaction patterns that are exhibited in human-autonomous agent teams and discusses future directions that can facilitate the design of human-compatible artificial agents.

CogSci 2021

Interaction Flexibility in Artificial Agents Teaming with Humans

### 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.

technical paper

Many real-world tasks require agents to coordinate their behavior to achieve shared goals. Successful collaboration requires not only adopting the same communicative conventions, but also grounding these conventions in the same task-appropriate conceptual abstractions. We investigate how humans use natural language to collaboratively solve physical assembly problems more effectively over time. Human participants were paired up in an online environment to reconstruct scenes containing two block towers. One participant could see the target towers, and sent assembly instructions for the other participant to reconstruct. Participants provided increasingly concise instructions across repeated attempts on each pair of towers, using more abstract referring expressions that captured each scene's hierarchical structure. To explain these findings, we extend recent probabilistic models of ad hoc convention formation with an explicit perceptual learning mechanism. These results shed light on the inductive biases that enable intelligent agents to coordinate upon shared procedural abstractions.

Learning to communicate about shared procedural abstractions

Primate cognition research allows us to reconstruct the evolution of human cognition. However, temporal and contextual factors that induce variation in cognitive studies with great apes are poorly understood. Here we report on a longitudinal study where we repeatedly tested a comparatively large sample of great apes (N = 40) with the same set of cognitive measures. We investigated the stability of group-level results, the reliability of individual differences, and the relation between cognitive performance and individual-level characteristics. We found results to be relatively stable on a group level. Some, but not all, tasks showed acceptable levels of reliability. Cognitive performance across tasks was not systematically related to any particular individual-level predictor. This study highlights the importance of methodological considerations – especially when studying individual differences – on the route to building a more robust science of primate cognitive evolution.

A Longitudinal Study of Great Ape Cognition: Stability, Reliability and the Influence of Individual Characteristics

In the face of strong evidence that a coin landed heads, can someone simply choose to believe it landed tails? Knowing that a large earthquake could result in personal tragedy, can someone simply choose to desire that it occur? We propose that in the face of strong reasons to adopt a given belief or desire, people are perceived to lack control: they cannot simply believe or desire otherwise. We test this “reason-based constraint” account of mental state change, and find that people reliably judge that evidence constrains belief formation, and utility constrains desire formation, in others. These results were not explained by a heuristic that simply treats irrational mental states as impossible to adopt intentionally. Rather, constraint results from the perceived influence of reasons on reasoning: people judge others as free to adopt irrational attitudes through actions that eliminate their awareness of strong reasons. These findings fill an important gap in our understanding of folk psychological reasoning, with implications for attributions of autonomy and moral responsibility.

Reason-Based Constraint in Theory of Mind

People can produce drawings of specific entities (e.g., Garfield), as well as general categories (e.g., ``cat''). What explains this ability to produce such varied drawings of even highly familiar object concepts? We hypothesized that drawing objects at different levels of abstraction depends on both sensory information and representational goals, such that drawings intended to portray a recently seen object preserve more detail than those intended to represent a category. Participants drew objects cued either with a photo or a category label. For each cue type, half the participants aimed to draw a specific exemplar; the other half aimed to draw the category. We found that label-cued category drawings were the most recognizable at the basic level, whereas photo-cued exemplar drawings were the least recognizable. Together, these findings highlight the importance of task context for explaining how people use drawings to communicate visual concepts in different ways.

Visual communication of object concepts at different levels of abstraction

There have been recently multiple calls to investigate the rhythmic behaviors (RBs) of nonhuman animals, as a way to gain insight into the evolution of human rhythm cognition and musicality. Currently, the empirical data from non-human species is scarce. Most strikingly, we lack data from chimpanzees (Pan troglodytes), our closest genetic relatives. Here, we present an observational study conducted at three sites (N=41), in which we systematically documented RBs in chimpanzees, with a particular focus on functional contexts, sex differences and emotional correlates. We found that RBs were frequent in chimpanzees, occurred primarily in social contexts, and often had social consequences. RBs were not exclusively associated with high arousal or playfulness. RBs were more frequent in males than females, but sex did not affect their social efficacy. Our findings are consistent with social theories on the evolution of musicality, but also highlight a role for RBs in chimpanzee inter-sexual communication.

Rhythmic behaviors in chimpanzees: range, functional contexts, sex differences and emotional correlates

Can people interpret communicative action modulations in terms of the actor’s distal goal? We investigated situations in which the proximal goal of an action (i.e., the movement endpoint) does not overlap with its distal goal (i.e., a final location beyond the movement endpoint). Participants were presented with animations of an object being moved at different velocities towards a designated endpoint. The distal goal, however, was for the object to be moved past this endpoint, to one of two occluded final locations. Participants were asked to select the location which they considered the likely distal goal of the action. As predicted, participants detected differences in movement velocity and, based on these differences, systematically mapped the movements to the two distal goal locations. These findings extend previous research on sensorimotor communication by demonstrating that communicative action modulations are not restricted to proximal goals but can also contain information about distal goals.

Understanding distal goals from proximal communicative actions

Paying attention to weight is important when deciding upon an object’s efficacy or value in various contexts (e.g. tool use, foraging). Proprioceptive discrimination learning, with objects that differ only in weight, has so far been investigated in a handful of primate species. Here we show that while Goffin’s cockatoos learn faster when additional colour cues are used, they can also quickly learn to discriminate between objects on the basis of their weight alone. Ultimately, the birds learned to discriminate between visually identical objects on the basis of weight much faster than primates, although methodological differences between tasks should be considered.

Goffin's cockatoos learn to discriminate objects based on weight alone in an object choice task

Speakers communicate to influence their partner's beliefs and shape their actions. Belief- and action-based objectives have been explored independently in recent computational models, but it has been challenging to explicitly compare or integrate them. Indeed, we find that they are conflated in standard referential communication tasks. To distinguish these accounts, we introduce a new paradigm called signaling bandits, generalizing classic Lewis signaling games to a multi-armed bandit setting where all targets in the context have some relative value. We develop three speaker models: a belief-oriented speaker with a purely informative objective; an action-oriented speaker with an instrumental objective; and a combined speaker which integrates the two by inducing listener beliefs that generally lead to desirable actions. We then present a series of simulations demonstrating that grounding production choices in future listener actions results in relevance effects and flexible uses of nonliteral language. More broadly, our findings suggest that language games based on richer decision problems are a promising avenue for insight into rational communication.

Extending rational models of communication from beliefs to actions

Planning, the process of evaluating the future consequences of actions, is typically formalized as search over a decision tree. This procedure increases expected rewards but is computationally expensive. Past attempts to understand how people mitigate the costs of planning have been guided by heuristics or the accumulation of prior experience, both of which are intractable in novel, high-complexity tasks. In this work, we propose a normative framework for optimizing the depth of tree search. Specifically, we model a metacognitive process via Bayesian inference to compute optimal planning depth. We show that our model makes sensible predictions over a range of parameters without relying on retrospection and that integrating past experiences into our model produces results that are consistent with the transition from goal-directed to habitual behavior over time and the uncertainty associated with prospective and retrospective estimates. Finally, we derive an online variant of our model that replicates these results.

Planning to plan: a Bayesian model for optimizing the depth of decision tree search

A large body of work has demonstrated the utility
of the Bayesian framework for capturing inference
in both specialist and everyday contexts. However,
the central tool of the framework, conditionalization
via Bayes’ rule, does not apply directly to a
common type of learning: the acquisition of conditional
information. How should an agent change
her beliefs on learning that “If A, then C”? This
issue, which is central to both reasoning and argumentation,
has recently prompted considerable research
interest. In this paper, we critique a prominent
proposal and provide a new, alternative, answer.

Interaction Flexibility in Artificial Agents Teaming with Humans

Downloads

Next from CogSci 2021

Learning to communicate about shared procedural abstractions

Similar lecture

Automatic computation of navigational affordances explains selective processing of geometry in scene perception: behavioral and computational evidence

Interaction Flexibility in Artificial Agents Teaming with Humans

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from CogSci 2021

Learning to communicate about shared procedural abstractions

Similar lecture

Automatic computation of navigational affordances explains selective processing of geometry in scene perception: behavioral and computational evidence

Downloads