United States

When inferring the goals that others are trying to achieve, people intuitively understand that others might make mistakes along the way. This is crucial for activities such as teaching, offering assistance, and deciding between blame or forgiveness. However, Bayesian models of theory of mind have generally not accounted for these mistakes, instead modeling agents as mostly optimal in achieving their goals. As a result, they are unable to explain phenomena like locking oneself out of one&#39;s house, or losing a game of chess. Here, we extend the Bayesian Theory of Mind framework to model boundedly rational agents who may have mistaken goals, plans, and actions. We formalize this by modeling agents as probabilistic programs, where goals may be confused with semantically similar states, plans may be misguided due to resource-bounded planning, and actions may be unintended due to execution errors. We present experiments eliciting human goal inferences in two domains: (i) a gridworld puzzle with gems locked behind doors, and (ii) a block-stacking domain. Our model better explains human inferences than alternatives, while generalizing across domains. These findings indicate the importance of modeling others as bounded agents, in order to account for the full richness of human intuitive psychology.

CogSci 2021

Modeling the Mistakes of Boundedly Rational Agents Within a Bayesian Theory of Mind

### 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.

When inferring the goals that others are trying to achieve, people intuitively understand that others might make mistakes along the way. This is crucial for activities such as teaching, offering assistance, and deciding between blame or forgiveness. However, Bayesian models of theory of mind have generally not accounted for these mistakes, instead modeling agents as mostly optimal in achieving their goals. As a result, they are unable to explain phenomena like locking oneself out of one's house, or losing a game of chess. Here, we extend the Bayesian Theory of Mind framework to model boundedly rational agents who may have mistaken goals, plans, and actions. We formalize this by modeling agents as probabilistic programs, where goals may be confused with semantically similar states, plans may be misguided due to resource-bounded planning, and actions may be unintended due to execution errors. We present experiments eliciting human goal inferences in two domains: (i) a gridworld puzzle with gems locked behind doors, and (ii) a block-stacking domain. Our model better explains human inferences than alternatives, while generalizing across domains. These findings indicate the importance of modeling others as bounded agents, in order to account for the full richness of human intuitive psychology.

technical paper

Liao et al. (2020) reported an eye-movement experiment in which subtitles were displayed at three different rates, with a key finding being that, with increasing speeds, participants made fewer, shorter fixations and longer saccades. To understand why these eye-movement behaviors might be adaptive, we completed simulations using the E-Z Reader model (Reichle et al., 2012) to examine how subtitle speed might affect word identification and sentence comprehension, as well as the efficacy of six possible compensatory reading strategies. These simulations suggest that the imposition of a lexical-processing deadline and/or strategy of skipping short words may support reading comprehension in impoverished conditions.

Using Simulations to Understand the Reading of Rapidly Displayed Subtitles

Children are often drawn to novelty, but these preferences may depend on their goals. In two experiments (N = 302), we show that children have differing preferences for novelty when seeking information compared to when they are asked to prioritize other goals. In Experiment 1, 4-7-year-olds wanted to have typical items (e.g., a four-legged chair) and learn about atypical items (e.g., a ten-legged chair). In Experiment 2, 4-6- year-olds wanted to learn about foreign characters, but liked foreign and local characters equally. We propose that children prefer to learn about novel instances for the promise of new information, which is evident in at least two domains (artifacts and people). However, this preference diminishes when children are asked about who they like, and it reverses to a familiarity preference when choosing between artifacts to acquire. In sum, our findings suggest that children’s preferences for novelty versus familiarity are sensitive to different goals.

Children's Novelty Preferences Depend on Information-Seeking Goals

It has been shown that complexity metrics, computed by a syntactic parser, is a predictor of human reading time, which is an approximation of human sentence comprehension difficulty. Nevertheless, parsers usually take as input sentences that have already been processed or even manually annotated. We propose to study a more realistic scenario, where the various processing levels (tokenization, PoS and morphology tagging, lemmatization, syntactic parsing and sentence segmentation) are predicted incrementally from raw text. To this end, we propose a versatile modeling framework, we call the Reading Machine, that performs all such linguistic tasks and allows to incorporate cognitive constrains such as incrementality. 
We illustrate the behavior of this setting through a case study where we test the hypothesis that the complexity metrics computed at different processing levels predicts human reading difficulty, and that when cognitive constraints are applied to the machine (e.g., incrementality), it yields better predictions.

On the Role of Low-level Linguistic Levels for Reading Time Prediction

This study investigated the extent to which varying interword spacing influences eye movement during reading in Arabic. Previous works conducted in Latin-script languages suggested that interword spaces facilitated word recognition. On the other hand, word recognition was inhibited when interword spaces were either removed or replaced by other characters (Rayner et al., 1998; Sheridan et al., 2013). We focused on the influence of interword spaces on reading Arabic which is characterized by the use of interword spaces and the position-informative allographic system. Based on an eye tracking experiment in which subjects read Arabic sentences presented in three levels of interword spacing and two levels of target word frequency, we found that eliminating interword spaces did not significantly inhibit reading, yet widening interword spaces exerted a facilitative effect. We argued that the effect of eliminating interword spaces was compensated by the ligating properties of Arabic letters during sentence reading, i.e. Arabic ligatures were position-informative which provided sufficient visual cues for word recognition regardless of the presence of interword spaces.

Eye movements when reading spaced and unspaced texts in Arabic

Representational Similarity Analysis (RSA) is a powerful tool for linking brain activity patterns to cognitive processes via similarity, allowing researchers to identify the neural substrates of different cognitive levels of representation. However, the ability to map between levels of representation and brain activity using similarity depends on underlying assumptions about the dynamics of cognitive processing. To demonstrate this point, we present three toy models that make different assumptions about the interactivity within the reading system, (1) discrete, feedforward, (2) cascading, feedforward and (3) fully interactive. With the temporal resolution of fMRI, only the discrete, feedforward model provides a straightforward mapping between activation similarity and level of representation. These simulations indicate the need for a cautious interpretation of RSA results, especially with processes that are highly interactive and with neuroimaging methods that have low temporal resolution. The study further suggests a role for fully-fleshed out computational models in RSA analyses.

Challenges for using Representational Similarity Analysis to Infer Cognitive Processes: A Demonstration from Interactive Activation Models of Word Reading

This study examines how linguistic knowledge is manifested in eye movements in reading, focusing on the effect of two key word properties: frequency and surprisal, on three progressively longer standard fixation measures: First Fixation, Gaze Duration and Total Fixation. Comparing English L1 speakers to a large and linguistically diverse group of English L2 speakers, we obtain the following results. 1) Word property effects on reading times are larger in L2 than in L1. 2) Differences between L1 and L2 speakers are substantially larger in the response to frequency than to surprisal. 3) The functional form of the relation between fixation times and frequency and surprisal in L2 is superlinear. 4) In L2 speakers, proficiency modulates frequency effects as a U shaped function. We discuss the implications of these results on theory of language processing and acquisition, as well as the general interpretation of frequency and surprisal effects in reading.

Eye Movement Traces of Linguistic Knowledge

We live in an age where robots are increasingly present in the social and moral world. Here, we explore how children and adults think about the mental lives and moral standing of robots. In Experiment 1 (N = 116), we found that children granted humans and robots with more mental life and vulnerability to harm than an anthropomorphized control (i.e., a toy bear). In Experiment 2 (N = 157), we found that, relative to children, adults ascribed less mental life and vulnerability to harm to robots. In Experiment 3 (N = 152), we modified our experiment to be within-subjects and measured beliefs concerning moral standing. Though younger children again appeared willing to assign mental capacities — particularly those related to experience (e.g., being capable of experiencing hunger) — to robots, older children and adults did so to a lesser degree. This diminished attribution of mental life tracked with diminished ratings of robot moral standing. This informs ongoing debates concerning emerging attitudes about artificial life.

Developmental changes in perceived moral standing of robots

We explore the impact of learning sequence on performance in a rule-plus-exception categorization task. Behavioural results indicate that exception categorization accuracy improves when exceptions are introduced later in learning, after exposure to rule-following stimuli. Simulations of this task using a neural network model of hippocampus and its subfields replicate these behavioural findings. Representational similarity analysis of the model’s hidden layers suggests that model representations are also impacted by trial sequence. These results provide novel computational evidence of hippocampus’s sensitivity to learning sequence and further support this region’s proposed role in category learning.

Learning exceptions to the rule in human and model via hippocampal encoding

This is a modern horror story about an innocently misbehaving
projector, and why we beseech everyone to report minimal col-
orimetric data about stimulus displays. We present anecdotal
experience of configuring a projector to display video stimuli
in a high-tesla MRI room, along with all the gotchas, (broken)
technical assumptions, and theoretical rehashings that should
be considered by every scientist who uses computers to dis-
play visual stimuli. The moral of our story is: (1) check that
your monitor/projector is actually showing the colors and lu-
minances that you think it is, (2) make explicit assumptions
regarding the physical/perceptual space of your stimuli and
how they relate to any model analysis you will perform. This
is especially important when modeling non-human animals,
since most equipment and data formats implicitly assume hu-
man perception. We show that innocent changes to display
settings such as brightness reduction can cause dangerously
unexpected results. Understanding and reporting colorimetric
data in scientific publications is important for two reasons: (1)
reproducibility, and (2) model fidelity.

Why We Should Report Colorimetric Data In Every Paper

The basal ganglia (BG) performs an important functional role in cognition, but models disagree about the nature of the relationship between BG activity and activity in other cortical areas. Previous computational models can be categorized as implementing the effects of the BG on prefrontal cortex as either local and direct, or involving other regions and, therefore, modulatory. To test which of these two effects best represents the role of the BG, a large fMRI dataset of 200 participants performing six, representative cognitive tasks was analyzed through Dynamic Causal Modeling (DCM). To ensure that DCM models were realistic and representative of a general brain architecture, the models were implemented within the putative neural underpinnings of the Common Model of Cognition, an abstract blueprint for cognition. The comparison showed that Mixed model, including both Direct and Modulatory connectivity, consistently outperformed models that included only direct or modulatory connections. It was also found that the relative rankings of the Direct and Modulatory models depended on the specific task, suggesting that the BG is a flexible system that adapts to task demands.

Modeling the Mistakes of Boundedly Rational Agents Within a Bayesian Theory of Mind

Downloads

Next from CogSci 2021

Using Simulations to Understand the Reading of Rapidly Displayed Subtitles

Similar lecture

Learning Evolved Combinatorial Symbols with a Neuro-symbolic Generative Model

Modeling the Mistakes of Boundedly Rational Agents Within a Bayesian Theory of Mind

.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

Using Simulations to Understand the Reading of Rapidly Displayed Subtitles

Similar lecture

Learning Evolved Combinatorial Symbols with a Neuro-symbolic Generative Model

Downloads