Skip to content

Commit

Permalink
added marple long cogsci paper
Browse files Browse the repository at this point in the history
  • Loading branch information
tobiasgerstenberg committed May 13, 2024
1 parent e760cfa commit 107defd
Show file tree
Hide file tree
Showing 32 changed files with 796 additions and 105 deletions.
4 changes: 2 additions & 2 deletions content/publication/cao2023semantics.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,10 @@

title = "A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models"
date = "2023-05-10"
authors = ["A. Cao","A. Geiger","E. Kreiss","T. Icard","T. Gerstenberg"]
authors = ["A. Cao\\*","A. Geiger\\*","E. Kreiss\\*","T. Icard","T. Gerstenberg"]
publication_types = ["3"]
publication_short = "_Proceedings of the 45th Annual Conference of the Cognitive Science Society_"
publication = "Cao A., Geiger A., Kreiss E., Icard T., Gerstenberg T. (2023). A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models. In _Proceedings of the 45th Annual Conference of the Cognitive Science Society_."
publication = "Cao A.\\*, Geiger A.\\*, Kreiss E.\\*, Icard T., Gerstenberg T. (2023). A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models. In _Proceedings of the 45th Annual Conference of the Cognitive Science Society_."
abstract = "When choosing how to describe what happened, we have a number of causal verbs at our disposal. In this paper, we develop a model-theoretic formal semantics for nine causal verbs that span the categories of CAUSE, ENABLE, and PREVENT. We use structural causal models (SCMs) to represent participants' mental construction of a scene when assessing the correctness of causal expressions relative to a presented context. Furthermore, SCMs enable us to model events relating both the physical world as well as agents' mental states. In experimental evaluations, we find that the proposed semantics exhibits a closer alignment with human evaluations in comparison to prior accounts of the verb families."
image_preview = ""
selected = false
Expand Down
4 changes: 2 additions & 2 deletions content/publication/gershman2016goals.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,10 +9,10 @@

title = "Plans, habits, and theory of mind"
date = "2016-01-01"
authors = ["S. J. Gershman ","T. Gerstenberg", "C. L. Baker", "F. Cushman"]
authors = ["S. J. Gershman","T. Gerstenberg", "C. L. Baker", "F. Cushman"]
publication_types = ["2"]
publication_short = "_PLoS ONE_"
publication = "Gershman*, S. J., Gerstenberg*, T., Baker, C. L., & Cushman, F. (2016). Plans, habits, and theory of mind. _PLoS ONE_, 11(9), e0162246. "
publication = "Gershman, S. J.\\*, Gerstenberg, T.\\*, Baker, C. L., & Cushman, F. (2016). Plans, habits, and theory of mind. _PLoS ONE_, 11(9), e0162246. "
image_preview = ""
selected = false
projects = []
Expand Down
4 changes: 2 additions & 2 deletions content/publication/mccoy2012probabilistic.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,10 +9,10 @@

title = "Why blame Bob? Probabilistic generative models, counterfactual reasoning, and blame attribution"
date = "2012-01-01"
authors = ["J. McCoy* ","T. D. Ullman* ","A. Stuhlmüller","T. Gerstenberg","J. B. Tenenbaum"]
authors = ["J. McCoy\\*","T. D. Ullman\\*","A. Stuhlmüller","T. Gerstenberg","J. B. Tenenbaum"]
publication_types = ["3"]
publication_short = "_Proceedings of the 34th Annual Conference of the Cognitive Science Society_"
publication = "McCoy, J.*, Ullman, T.*, Stuhlmüller, A., Gerstenberg, T., & Tenenbaum, J. B. (2012). Why blame Bob? Probabilistic generative models, counterfactual reasoning, and blame attribution. In _Proceedings of the 34th Annual Conference of the Cognitive Science Society_ (pp. 1996-2001). Austin, TX: Cognitive Science Society."
publication = "McCoy, J.\\*, Ullman, T.\\*, Stuhlmüller, A., Gerstenberg, T., & Tenenbaum, J. B. (2012). Why blame Bob? Probabilistic generative models, counterfactual reasoning, and blame attribution. In _Proceedings of the 34th Annual Conference of the Cognitive Science Society_ (pp. 1996-2001). Austin, TX: Cognitive Science Society."
image_preview = ""
selected = false
projects = []
Expand Down
4 changes: 2 additions & 2 deletions content/publication/outa2022stop.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,10 @@

title = "Stop, children what's that sound? Multi-modal inference through mental simulation"
date = "2022-05-12"
authors = ["J. Outa","X. Jia Zhou","H. Gweon","T. Gerstenberg"]
authors = ["J. Outa\\*","X. Jia Zhou\\*","H. Gweon","T. Gerstenberg"]
publication_types = ["3"]
publication_short = "_Cognitive Science Proceedings_"
publication = "Outa* J., Zhou* X. J., Gweon H., Gerstenberg T. (2022). Stop, children what's that sound? Multi-modal inference through mental simulation. In _Cognitive Science Proceedings_."
publication = "Outa J.\\*, Zhou X. J.\\*, Gweon H., Gerstenberg T. (2022). Stop, children what's that sound? Multi-modal inference through mental simulation. In _Cognitive Science Proceedings_."
image_preview = ""
abstract = "Human adults can figure out what happened by combining evidence from different sensory modalities, such as vision and sound. How does the ability to integrate multi-modal information develop in early childhood? Inspired by prior computational work and behavioral studies with adults, we examined 3- to 8-year-old children's ability to reason about the physical trajectory of a ball that was dropped into an occluded Plinko box. Children had to infer in which one of three holes the ball was dropped based on visual information (i.e., where the ball landed) and auditory information (i.e., the sounds of the ball colliding with parts of the box). We compare children's responses to the predictions of four computational models. The results suggest that although even the youngest children make systematic judgments rather than randomly guessing, children's ability to integrate visual and auditory evidence continues to develop into late childhood."
selected = false
Expand Down
33 changes: 33 additions & 0 deletions content/publication/wu2024whodunnit.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
+++
# 0 -> 'Forthcoming',
# 1 -> 'Preprint',
# 2 -> 'Journal',
# 3 -> 'Conference Proceedings',
# 4 -> 'Book chapter',
# 5 -> 'Thesis'

title = "Whodunnit? Inferring what happened from multimodal evidence"
date = "2024-05-13"
authors = ["S. A Wu\\*","E. Brockbank\\*","H. Cha","J. Fränken","E. Jin","Z. Huang","W. Liu","R. Zhang","J. Wu","T. Gerstenberg"]
publication_types = ["3"]
publication_short = "_Proceedings of the 46th Annual Conference of the Cognitive Science Society_"
publication = "Wu*, S. A., Brockbank*, E., Cha, H., Fränken, J., Jin, E., Huang, Z., Liu, W., Zhang, R., Wu, J., Gerstenberg, T. (2024). Whodunnit? Inferring what happened from multimodal evidence. In _Proceedings of the 46th Annual Conference of the Cognitive Science Society_."
abstract = "Humans are remarkably adept at inferring the causes of events in their environment; doing so often requires incorporating information from multiple sensory modalities. For instance, if a car slows down in front of us, inferences about why they did so are rapidly revised if we also hear sirens in the distance. Here, we investigate the ability to reconstruct others' actions and events from the past by integrating multimodal information. Participants were asked to infer which of two agents performed an action in a household setting given either visual evidence, auditory evidence, or both. We develop a computational model that makes inferences by generating multimodal simulations, and also evaluate our task on a large language model (GPT-4) and a large multimodal model (GPT-4V). We find that humans are relatively accurate overall and perform best when given multimodal evidence. GPT-4 and GPT-4V performance comes close overall, but is very weakly correlated with participants across individual trials. Meanwhile, the simulation model captures the pattern of human responses well. Multimodal event reconstruction represents a challenge for current AI systems, and frameworks that draw on the cognitive processes underlying people's ability to reconstruct events offer a promising avenue forward."
image_preview = ""
selected = false
projects = []
url_pdf = "papers/wu2024whodunnit.pdf"
url_preprint = ""
url_code = ""
url_dataset = ""
url_slides = ""
url_video = ""
url_poster = ""
url_source = ""
url_custom = [{name = "Github", url = "https://github.com/cicl-stanford/whodunnit_multimodal_inference"}]
math = true
highlight = true
[header]
# image = "publications/wu2024whodunnit.png"
caption = ""
+++
4 changes: 2 additions & 2 deletions content/publication/zhang2023llm.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,10 @@

title = "You are what you're for: Essentialist categorization in large language models"
date = "2023-05-11"
authors = ["S. Zhang","J. S. She","T. Gerstenberg","D. Rose"]
authors = ["S. Zhang\\*","J. S. She\\*","T. Gerstenberg","D. Rose"]
publication_types = ["3"]
publication_short = "_Proceedings of the 45th Annual Conference of the Cognitive Science Society_"
publication = "Zhang S., She J. S., Gerstenberg T., Rose D. (2023). You are what you're for: Essentialist categorization in large language models. In _Proceedings of the 45th Annual Conference of the Cognitive Science Society_."
publication = "Zhang S.\\*, She J. S.\\*, Gerstenberg T., Rose D. (2023). You are what you're for: Essentialist categorization in large language models. In _Proceedings of the 45th Annual Conference of the Cognitive Science Society_."
abstract = "How do essentialist beliefs about categories arise? We hypothesize that such beliefs are transmitted via language. We subject large language models (LLMs) to vignettes from the literature on essentialist categorization and find that they align well with people when the studies manipulated teleological information - information about what something is for. We examine whether in a classic test of essentialist categorization - the transformation task - LLMs prioritize teleological properties over information about what something looks like, or is made of. Experiments 1 and 2 find that telos and what something is made of matter more than appearance. Experiment 3 manipulates all three factors and finds that what something is for matters more than what it's made of. Overall, these studies suggest that language alone may be sufficient to give rise to essentialist beliefs, and that information about what something is for matters more."
image_preview = ""
selected = false
Expand Down
8 changes: 4 additions & 4 deletions docs/404.html
Original file line number Diff line number Diff line change
Expand Up @@ -237,6 +237,10 @@ <h1>Page not found</h1>

<h2>Publications</h2>

<ul>
<li><a href="https://cicl.stanford.edu/publication/wu2024whodunnit/">Whodunnit? Inferring what happened from multimodal evidence</a></li>
</ul>

<ul>
<li><a href="https://cicl.stanford.edu/publication/tsirtsis2024sequential/">Towards a computational model of responsibility judgments in sequential human-AI collaboration</a></li>
</ul>
Expand All @@ -253,10 +257,6 @@ <h2>Publications</h2>
<li><a href="https://cicl.stanford.edu/publication/kirfel2024do/">Do as I explain: Explanations communicate optimal interventions</a></li>
</ul>

<ul>
<li><a href="https://cicl.stanford.edu/publication/keshmirian2024biased/">Biased causal strength judgments in humans and large language models</a></li>
</ul>




Expand Down
12 changes: 11 additions & 1 deletion docs/bibtex/cic_papers.bib
Original file line number Diff line number Diff line change
@@ -1,13 +1,23 @@
%% This BibTeX bibliography file was created using BibDesk.
%% https://bibdesk.sourceforge.io/
%% Created for Tobias Gerstenberg at 2024-05-12 17:48:57 +0200
%% Created for Tobias Gerstenberg at 2024-05-13 09:25:20 +0200
%% Saved with string encoding Unicode (UTF-8)
@inproceedings{wu2024whodunnit,
abstract = {Humans are remarkably adept at inferring the causes of events in their environment; doing so often requires incorporating information from multiple sensory modalities. For instance, if a car slows down in front of us, inferences about why they did so are rapidly revised if we also hear sirens in the distance. Here, we investigate the ability to reconstruct others' actions and events from the past by integrating multimodal information. Participants were asked to infer which of two agents performed an action in a household setting given either visual evidence, auditory evidence, or both. We develop a computational model that makes inferences by generating multimodal simulations, and also evaluate our task on a large language model (GPT-4) and a large multimodal model (GPT-4V). We find that humans are relatively accurate overall and perform best when given multimodal evidence. GPT-4 and GPT-4V performance comes close overall, but is very weakly correlated with participants across individual trials. Meanwhile, the simulation model captures the pattern of human responses well. Multimodal event reconstruction represents a challenge for current AI systems, and frameworks that draw on the cognitive processes underlying people's ability to reconstruct events offer a promising avenue forward.},
author = {Wu, Sarah A and Brockbank, Erik and Cha, Hannah and Fr{\"a}nken, Jan-Philipp and Jin, Emily and Huang, Zhuoyi and Liu, Weiyu and Zhang, Ruohan and Wu, Jiajun and Gerstenberg, Tobias},
booktitle = {{Proceedings of the 46th Annual Conference of the Cognitive Science Society}},
date-added = {2024-05-13 09:25:14 +0200},
date-modified = {2024-05-13 09:25:14 +0200},
editor = {Larissa K Samuelson and Stefan Frank and Mariya Toneva and Allyson Mackey and Eliot Hazeltine},
title = {{Whodunnit? Inferring what happened from multimodal evidence}},
year = {2024}}

@inproceedings{tsirtsis2024sequential,
abstract = {When a human and an AI agent collaborate to complete a task and something goes wrong, who is responsible? Prior work has developed theories to describe how people assign responsibility to individuals in teams. However, there has been little work studying the cognitive processes that underlie responsibility judgments in human-AI collaborations, especially for tasks comprising a sequence of interdependent actions. In this work, we take a step towards filling this gap. Using semiautonomous driving as a paradigm, we develop an environment that simulates stylized cases of human-AI collaboration using a generative model of agent behavior. We propose a model of responsibility that considers how unexpected an agent's action was, and what would have happened had they acted differently. We test the model's predictions empirically and find that in addition to action expectations and counterfactual considerations, participants' responsibility judgments are also affected by how much each agent actually contributed to the outcome.},
author = {Tsirtsis, Stratis and Gomez-Rodriguez, Manuel and Gerstenberg, Tobias},
Expand Down
2 changes: 1 addition & 1 deletion docs/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,7 @@
<meta property="og:description" content="">
<meta property="og:locale" content="en-us">

<meta property="og:updated_time" content="2024-05-12T00:00:00&#43;00:00">
<meta property="og:updated_time" content="2024-05-13T00:00:00&#43;00:00">



Expand Down
20 changes: 10 additions & 10 deletions docs/index.xml
Original file line number Diff line number Diff line change
Expand Up @@ -6,9 +6,18 @@
<generator>Hugo -- gohugo.io</generator>
<language>en-us</language>
<copyright>&amp;copy; 2024 Tobias Gerstenberg</copyright>
<lastBuildDate>Sun, 12 May 2024 00:00:00 +0000</lastBuildDate>
<lastBuildDate>Mon, 13 May 2024 00:00:00 +0000</lastBuildDate>
<atom:link href="/" rel="self" type="application/rss+xml" />

<item>
<title>Whodunnit? Inferring what happened from multimodal evidence</title>
<link>https://cicl.stanford.edu/publication/wu2024whodunnit/</link>
<pubDate>Mon, 13 May 2024 00:00:00 +0000</pubDate>

<guid>https://cicl.stanford.edu/publication/wu2024whodunnit/</guid>
<description></description>
</item>

<item>
<title>Towards a computational model of responsibility judgments in sequential human-AI collaboration</title>
<link>https://cicl.stanford.edu/publication/tsirtsis2024sequential/</link>
Expand Down Expand Up @@ -135,14 +144,5 @@
<description></description>
</item>

<item>
<title>MoCa: Measuring human-language model alignment on causal and moral judgment tasks</title>
<link>https://cicl.stanford.edu/publication/nie2023moca/</link>
<pubDate>Fri, 20 Oct 2023 00:00:00 +0000</pubDate>

<guid>https://cicl.stanford.edu/publication/nie2023moca/</guid>
<description></description>
</item>

</channel>
</rss>
53 changes: 48 additions & 5 deletions docs/member/tobias_gerstenberg/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -356,6 +356,49 @@ <h2 id="publications">Publications</h2>


<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
S. A Wu*, E. Brockbank*, H. Cha, J. Fränken, E. Jin, Z. Huang, W. Liu, R. Zhang, J. Wu, T. Gerstenberg</span>

(2024).

<a href="https://cicl.stanford.edu/publication/wu2024whodunnit/" itemprop="name">Whodunnit? Inferring what happened from multimodal evidence</a>.
<em>Proceedings of the 46th Annual Conference of the Cognitive Science Society</em>.




<p>





<a class="btn btn-outline-primary my-1 mr-1 btn-sm" href="https://cicl.stanford.edu/papers/wu2024whodunnit.pdf" target="_blank" rel="noopener">
PDF
</a>














<a class="btn btn-outline-primary my-1 mr-1 btn-sm" href="https://github.com/cicl-stanford/whodunnit_multimodal_inference" target="_blank" rel="noopener">
Github
</a>


</p>

</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
S. Tsirtsis, M. Gomez-Rodriguez, T. Gerstenberg</span>

Expand Down Expand Up @@ -1463,7 +1506,7 @@ <h2 id="publications">Publications</h2>
</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
S. Zhang, J. S. She, T. Gerstenberg, D. Rose</span>
S. Zhang*, J. S. She*, T. Gerstenberg, D. Rose</span>

(2023).

Expand Down Expand Up @@ -1514,7 +1557,7 @@ <h2 id="publications">Publications</h2>
</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
A. Cao, A. Geiger, E. Kreiss, T. Icard, T. Gerstenberg</span>
A. Cao*, A. Geiger*, E. Kreiss*, T. Icard, T. Gerstenberg</span>

(2023).

Expand Down Expand Up @@ -2002,7 +2045,7 @@ <h2 id="publications">Publications</h2>
</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
J. Outa, X. Jia Zhou, H. Gweon, T. Gerstenberg</span>
J. Outa*, X. Jia Zhou*, H. Gweon, T. Gerstenberg</span>

(2022).

Expand Down Expand Up @@ -3781,7 +3824,7 @@ <h2 id="publications">Publications</h2>
</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
S. J. Gershman , T. Gerstenberg, C. L. Baker, F. Cushman</span>
S. J. Gershman, T. Gerstenberg, C. L. Baker, F. Cushman</span>

(2016).

Expand Down Expand Up @@ -4759,7 +4802,7 @@ <h2 id="publications">Publications</h2>
</div>
<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
<span itemprop="author">
J. McCoy* , T. D. Ullman* , A. Stuhlmüller, T. Gerstenberg, J. B. Tenenbaum</span>
J. McCoy*, T. D. Ullman*, A. Stuhlmüller, T. Gerstenberg, J. B. Tenenbaum</span>

(2012).

Expand Down
Binary file added docs/papers/wu2024whodunnit.pdf
Binary file not shown.
4 changes: 2 additions & 2 deletions docs/publication/cao2023semantics/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -239,7 +239,7 @@
<h1 itemprop="name" class ="title-text">A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models</h1>
<p class="pub-authors" itemprop="author">

A. Cao, A. Geiger, E. Kreiss, T. Icard, T. Gerstenberg
A. Cao*, A. Geiger*, E. Kreiss*, T. Icard, T. Gerstenberg

</p>
<span class="pull-right">
Expand Down Expand Up @@ -283,7 +283,7 @@ <h3>Abstract</h3>
<div class="col-sm-10">
<div class="row">
<div class="col-xs-12 col-sm-3 pub-row-heading">Publication</div>
<div class="col-xs-12 col-sm-9">Cao A., Geiger A., Kreiss E., Icard T., Gerstenberg T. (2023). A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models. In <em>Proceedings of the 45th Annual Conference of the Cognitive Science Society</em>.</div>
<div class="col-xs-12 col-sm-9">Cao A.*, Geiger A.*, Kreiss E.*, Icard T., Gerstenberg T. (2023). A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models. In <em>Proceedings of the 45th Annual Conference of the Cognitive Science Society</em>.</div>
</div>
</div>
<div class="col-sm-1"></div>
Expand Down
Loading

0 comments on commit 107defd

Please sign in to comment.