Quantifying Humans’ Priors Over Graphical Representations of Tasks

Gecia Bravo Hermsdorff, Talmo Pereira, Yael Niv

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

Some new tasks are trivial to learn while others are almost impossible; what determines how easy it is to learn an arbitrary task? Similar to how our prior beliefs about new visual scenes colors our perception of new stimuli, our priors about the structure of new tasks shapes our learning and generalization abilities [2]. While quantifying visual priors has led to major insights on how our visual system works [5, 10, 11], quantifying priors over tasks remains a formidable goal, as it is not even clear how to define a task [4]. Here, we focus on tasks that have a natural mapping to graphs. We develop a method to quantify humans’ priors over these “task graphs”, combining new modeling approaches with Markov chain Monte Carlo with people, MCMCP (a process whereby an agent learns from data generated by another agent, recursively [9]). We show that our method recovers priors more accurately than a standard MCMC sampling approach. Additionally, we propose a novel low-dimensional “smooth” (In the sense that graphs that differ by fewer edges are given similar probabilities.) parametrization of probability distributions over graphs that allows for more accurate recovery of the prior and better generalization. We have also created an online experiment platform that gamifies our MCMCP algorithm and allows subjects to interactively draw the task graphs. We use this platform to collect human data on several navigation and social interactions tasks. We show that priors over these tasks have non-trivial structure, deviating significantly from null models that are insensitive to the graphical information. The priors also notably differ between the navigation and social domains, showing fewer differences between cover stories within the same domain. Finally, we extend our framework to the more general case of quantifying priors over exchangeable random structures.

Original languageEnglish (US)
Title of host publicationSpringer Proceedings in Complexity
PublisherSpringer
Pages281-290
Number of pages10
DOIs
StatePublished - Jan 1 2018

Publication series

NameSpringer Proceedings in Complexity

Fingerprint

Navigation
Markov processes
Probability distributions
Sampling
Color
Recovery
Experiments

All Science Journal Classification (ASJC) codes

  • Applied Mathematics
  • Computer Science Applications
  • Modeling and Simulation

Cite this

Hermsdorff, G. B., Pereira, T., & Niv, Y. (2018). Quantifying Humans’ Priors Over Graphical Representations of Tasks. In Springer Proceedings in Complexity (pp. 281-290). (Springer Proceedings in Complexity). Springer. https://doi.org/10.1007/978-3-319-96661-8_30
Hermsdorff, Gecia Bravo ; Pereira, Talmo ; Niv, Yael. / Quantifying Humans’ Priors Over Graphical Representations of Tasks. Springer Proceedings in Complexity. Springer, 2018. pp. 281-290 (Springer Proceedings in Complexity).
@inbook{6788d344a1104997959fb38cda58cdcb,
title = "Quantifying Humans’ Priors Over Graphical Representations of Tasks",
abstract = "Some new tasks are trivial to learn while others are almost impossible; what determines how easy it is to learn an arbitrary task? Similar to how our prior beliefs about new visual scenes colors our perception of new stimuli, our priors about the structure of new tasks shapes our learning and generalization abilities [2]. While quantifying visual priors has led to major insights on how our visual system works [5, 10, 11], quantifying priors over tasks remains a formidable goal, as it is not even clear how to define a task [4]. Here, we focus on tasks that have a natural mapping to graphs. We develop a method to quantify humans’ priors over these “task graphs”, combining new modeling approaches with Markov chain Monte Carlo with people, MCMCP (a process whereby an agent learns from data generated by another agent, recursively [9]). We show that our method recovers priors more accurately than a standard MCMC sampling approach. Additionally, we propose a novel low-dimensional “smooth” (In the sense that graphs that differ by fewer edges are given similar probabilities.) parametrization of probability distributions over graphs that allows for more accurate recovery of the prior and better generalization. We have also created an online experiment platform that gamifies our MCMCP algorithm and allows subjects to interactively draw the task graphs. We use this platform to collect human data on several navigation and social interactions tasks. We show that priors over these tasks have non-trivial structure, deviating significantly from null models that are insensitive to the graphical information. The priors also notably differ between the navigation and social domains, showing fewer differences between cover stories within the same domain. Finally, we extend our framework to the more general case of quantifying priors over exchangeable random structures.",
author = "Hermsdorff, {Gecia Bravo} and Talmo Pereira and Yael Niv",
year = "2018",
month = "1",
day = "1",
doi = "https://doi.org/10.1007/978-3-319-96661-8_30",
language = "English (US)",
series = "Springer Proceedings in Complexity",
publisher = "Springer",
pages = "281--290",
booktitle = "Springer Proceedings in Complexity",

}

Hermsdorff, GB, Pereira, T & Niv, Y 2018, Quantifying Humans’ Priors Over Graphical Representations of Tasks. in Springer Proceedings in Complexity. Springer Proceedings in Complexity, Springer, pp. 281-290. https://doi.org/10.1007/978-3-319-96661-8_30

Quantifying Humans’ Priors Over Graphical Representations of Tasks. / Hermsdorff, Gecia Bravo; Pereira, Talmo; Niv, Yael.

Springer Proceedings in Complexity. Springer, 2018. p. 281-290 (Springer Proceedings in Complexity).

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Quantifying Humans’ Priors Over Graphical Representations of Tasks

AU - Hermsdorff, Gecia Bravo

AU - Pereira, Talmo

AU - Niv, Yael

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Some new tasks are trivial to learn while others are almost impossible; what determines how easy it is to learn an arbitrary task? Similar to how our prior beliefs about new visual scenes colors our perception of new stimuli, our priors about the structure of new tasks shapes our learning and generalization abilities [2]. While quantifying visual priors has led to major insights on how our visual system works [5, 10, 11], quantifying priors over tasks remains a formidable goal, as it is not even clear how to define a task [4]. Here, we focus on tasks that have a natural mapping to graphs. We develop a method to quantify humans’ priors over these “task graphs”, combining new modeling approaches with Markov chain Monte Carlo with people, MCMCP (a process whereby an agent learns from data generated by another agent, recursively [9]). We show that our method recovers priors more accurately than a standard MCMC sampling approach. Additionally, we propose a novel low-dimensional “smooth” (In the sense that graphs that differ by fewer edges are given similar probabilities.) parametrization of probability distributions over graphs that allows for more accurate recovery of the prior and better generalization. We have also created an online experiment platform that gamifies our MCMCP algorithm and allows subjects to interactively draw the task graphs. We use this platform to collect human data on several navigation and social interactions tasks. We show that priors over these tasks have non-trivial structure, deviating significantly from null models that are insensitive to the graphical information. The priors also notably differ between the navigation and social domains, showing fewer differences between cover stories within the same domain. Finally, we extend our framework to the more general case of quantifying priors over exchangeable random structures.

AB - Some new tasks are trivial to learn while others are almost impossible; what determines how easy it is to learn an arbitrary task? Similar to how our prior beliefs about new visual scenes colors our perception of new stimuli, our priors about the structure of new tasks shapes our learning and generalization abilities [2]. While quantifying visual priors has led to major insights on how our visual system works [5, 10, 11], quantifying priors over tasks remains a formidable goal, as it is not even clear how to define a task [4]. Here, we focus on tasks that have a natural mapping to graphs. We develop a method to quantify humans’ priors over these “task graphs”, combining new modeling approaches with Markov chain Monte Carlo with people, MCMCP (a process whereby an agent learns from data generated by another agent, recursively [9]). We show that our method recovers priors more accurately than a standard MCMC sampling approach. Additionally, we propose a novel low-dimensional “smooth” (In the sense that graphs that differ by fewer edges are given similar probabilities.) parametrization of probability distributions over graphs that allows for more accurate recovery of the prior and better generalization. We have also created an online experiment platform that gamifies our MCMCP algorithm and allows subjects to interactively draw the task graphs. We use this platform to collect human data on several navigation and social interactions tasks. We show that priors over these tasks have non-trivial structure, deviating significantly from null models that are insensitive to the graphical information. The priors also notably differ between the navigation and social domains, showing fewer differences between cover stories within the same domain. Finally, we extend our framework to the more general case of quantifying priors over exchangeable random structures.

UR - http://www.scopus.com/inward/record.url?scp=85051738688&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051738688&partnerID=8YFLogxK

U2 - https://doi.org/10.1007/978-3-319-96661-8_30

DO - https://doi.org/10.1007/978-3-319-96661-8_30

M3 - Chapter

T3 - Springer Proceedings in Complexity

SP - 281

EP - 290

BT - Springer Proceedings in Complexity

PB - Springer

ER -

Hermsdorff GB, Pereira T, Niv Y. Quantifying Humans’ Priors Over Graphical Representations of Tasks. In Springer Proceedings in Complexity. Springer. 2018. p. 281-290. (Springer Proceedings in Complexity). https://doi.org/10.1007/978-3-319-96661-8_30