Causal Structure Learning with Continuous Variables in Continuous Time

Interventions, time, and continuous-valued variables are all potentially powerful cues to causation. Furthermore, when observed over time, causal processes can contain feedback and oscillatory dynamics that make inference hard. We present a generative model and framework for causal inference over continuous variables in continuous time based on Ornstein-Uhlenbeck processes. Our generative model produces a stochastic sequence of evolving variable values that manifest many dynamical properties depending on the nature of the causal relationships, and a learner’s interventions (manual changes to the values of variables during a trial). Our model is also invertible, allowing us to benchmark participant judgments against an optimal model. We find that when interacting with systems acting according to this formalism people directly compare relationships between individual variable pairs rather than considering the full space of possible models, in accordance with a local computations model of causal learning (e.g., Fernbach & Sloman, 2009). The formalism presented here provides researchers in causal cognition with a powerful framework for studying dynamic systems and presents opportunities for other areas in cognitive psychology such as control problems.

[1]  Thomas L. Griffiths,et al.  A rational model of causal induction with continuous causes , 2011, NIPS 2011.

[2]  P. Cheng.  From covariation to causation: A causal power theory. , 1997 .

[3]  Joshua B. Tenenbaum,et al.  Inferring causal networks from observations and interventions , 2003, Cogn. Sci..

[4]  A. Yuille,et al.  Bayesian generic priors for causal learning. , 2008, Psychological review.

[5]  B. Rehder.  Reasoning With Causal Cycles. , 2017, Cognitive science.

[6]  D. Broadbent,et al.  On the Relationship between Task Performance and Associated Verbalizable Knowledge , 1984 .

[7]  Philip M. Fernbach,et al.  Causal learning with local computations. , 2009, Journal of experimental psychology. Learning, memory, and cognition.

[8]  David A. Lagnado,et al.  The order of things: Inferring causal structure from temporal patterns , 2014, CogSci.

[9]  Michael I. Jordan,et al.  Forward Models: Supervised Learning with a Distal Teacher , 1992, Cogn. Sci..

[10]  N. Shephard,et al.  Non‐Gaussian Ornstein–Uhlenbeck‐based models and some of their uses in financial economics , 2001 .

[11]  Tom L. Beauchamp,et al.  An enquiry concerning human understanding : a critical edition , 2000 .

[12]  Joshua B Tenenbaum,et al.  Theory-based causal induction. , 2009, Psychological review.

[13]  G. Uhlenbeck,et al.  On the Theory of the Brownian Motion , 1930 .

[14]  Kevin Soo,et al.  Causal Learning With Continuous Variables Over Time , 2016, CogSci.

[15]  B. Rehder,et al.  Strategies to intervene on causal systems are adaptively selected , 2015, Cognitive Psychology.

[16]  Todd M. Gureckis,et al.  CUNY Academic , 2016 .

[17]  David A. Lagnado,et al.  Causal learning from interventions and dynamics in continuous time , 2017, CogSci.

[18]  David M. Sobel,et al.  A theory of causal learning in children: causal maps and Bayes nets. , 2004, Psychological review.

[19]  Thomas L. Griffiths,et al.  Formalizing Neurath’s Ship: Approximate Algorithms for Online Causal Learning , 2016, Psychological review.

[20]  J. Tenenbaum,et al.  Structure and strength in causal induction , 2005, Cognitive Psychology.

[21]  Vladimír Lacko,et al.  Planning of experiments for a nonautonomous ornstein-uhlenbeck process , 2012 .

[22]  Jessica B. Hamrick,et al.  psiTurk: An open-source framework for conducting replicable behavioral experiments online , 2016, Behavior research methods.

[23]  Thomas L. Griffiths,et al.  Upsetting the contingency table: Causal induction over sequences of point events , 2015, CogSci.

[24]  M. Buehner,et al.  Temporal predictability facilitates causal learning. , 2010, Journal of experimental psychology. General.

[25]  George A. Alvarez,et al.  Explaining human multiple object tracking as resource-constrained approximate inference in a dynamic probabilistic model , 2009, NIPS.