The Limits Of Normality Effects On Causal Judgments

Evidence From Continuous Causes

Kaylee Miceli, Nina Van Rooy, Felipe De Brigard

Department of Psychology & Neuroscience, Department of Philosophy
Duke University

Kevin O’Neill

Department of Experimental Psychology, Institute of Cognitive Neuroscience
University College London

A familiar example

Icard et al. (2017)

A formal description

\(C=1\), \(A=1\).

So, \(E = 1\).

To what extent did \(C=1\) cause \(E=1\)?

Two causal structures

Conjunctive

\(E = \; C \wedge A\)

  • \(C=1\) and \(A=1\) are both necessary
  • Neither \(C=1\) nor \(A=1\) is sufficient

Disjunctive

\(E = \; C \vee A\)

  • Neither \(C=1\) nor \(A=1\) are necessary
  • Either \(C=1\) or \(A=1\) is sufficient

Icard et al. (2017)

A slightly more realistic example

A formal description

\(C=c \;=\; A=a\).

\(c+a > \theta\), so \(E = 1\).

To what extent did \(C=c\) cause \(E=1\)?

Two special cases

Conjunctive

\(c+a=\theta+1\)

  • \(C\ge c\) and \(A\ge a\) are both necessary
  • Neither \(C=c\) nor \(A=c\) is sufficient

Disjunctive

\(c=a=\theta+1\)

  • Neither \(C\ge c\) nor \(A\ge a\) are necessary
  • Either \(C=c\) or \(A=a\) is sufficient



Do norms affect causal judgments?

Design

Experiment 1

Manipulations

Participants learned the distribution of events

Participants found abnormal events surprising

No influence of norms on
causal judgments

Experiment 2

Manipulations

Participants learned the distribution of events

Participants found abnormal events surprising

No influence of norms on
causal judgments

Conclusions

Conclusions

  • Actual causation is well-studied when events are binary
  • But we don’t know much about cases where causes are continuous variables!
  • Even when participants learn relevant norms, they don’t affect causal judgments in such contexts

thanks!

References

Icard, Thomas F, Jonathan F Kominsky, and Joshua Knobe. 2017. “Normality and Actual Causal Strength.” Cognition 161: 80–93.