The Neural Basis of Belief Encoding and Integration in Moral Judgment Liane Young & Rebecca Saxe
Goals of the Article
Examine the role of belief attribution in moral judgment
Investigate the neural correlates of belief encoding and belief integration during moral judgment.
Are there distinct cognitive and neural processes for encoding and integration of belief information?
Theory of Mind
The ability to attribute mental states (beliefs, desires, etc) to others.
Mental states are an important component in the formation of a moral judgments.
Intentional harms are judged as significantly worse than accidental harms.
This is nicely mirrored in the legal system’s concept of mens rea.
Neural Basis of Theory of Mind
In multiple studies, ToM has been found to correlate with a consistent set of brain regions.
Medial Prefrontal Cortex (MPFC)
Right and Left Temporo-Parietal Junction (TPJ)
The RTPJ appears to be particularly selective for belief attribution.
Encoding vs Integrating Belief Attributions
By age 4, most children are able to pass a “false belief” task.
However, in moral judgment tasks, they do not use this information in the formation of moral judgments until several years later.
For example, 4 year olds judge a person who intends to harm another, but accidentally helps them as much worse than a person who intends to help someone, but accidentally harms them.
Encoding vs Integrating
This suggests that there may be separate cognitive processes for the encoding of mental state information (the ability to represent another person’s mental state) and the integration of mental state information (the ability to use mental state information in conjunction with other information when making judgments and decisions).
1. Brain regions used during encoding should be:
active when belief information is initially presented.
Selective for belief over non-belief information.
2. Brain regions for integration should be:
Recruited when morally relevant non-belief (outcome) information is presented.
Show an interaction between belief and outcome information.
Participants read 48 vignettes in a 2x2x2 design.
Outcome (negative or neutral)
Belief (negative or neutral)
Order (belief or foreshadowing first)
Vignettes were presented in four parts over 24s.
After presentation of each vignette, Ps make a moral judgment about whether an action was permissible or forbidden.
In addition to the primary task, participants completed a localizer task.
Ps made inferences that either did or did not clearly required belief attribution.
Localizer tasks are used to functionally define a region of interest in the brain.
There are significant pros and cons to using a localizer approach.
Actions performed by agents with negative beliefs were significantly less permissible than those performed by agents with neutral beliefs.
Actions that led to negative outcomes were significantly less permissible than actions that led to neutral outcomes.
No interaction between belief and outcome.
Analyses were run at three different time points:
Time 1: Belief or Foreshadow (10-14s)
Time 2: Foreshadow or Belief (16-20s)
Time 3: Action and Outcome (22-26s)
Times 1 and 2 represent encoding phases, while time 3 represents the integration phase.
Significant time (1 or 2) x order (belief or foreshadow first) interaction.
Activation in RTPJ appears to track belief information.
Encoding Results Similar, though less robust, results were found in the PC and LTPJ.
No significant effects during encoding phase in dMPFC.
Time 3: Agent’s action and the outcome of vignette are revealed.
Analyzed using a 2x2x2 ANOVA
Outcome (negative vs. neutral)
Belief (negative vs. neutral)
Order (belief first vs foreshadow first)
Integration Results (Foreshadow First)
RTPJ: Significant main effect of belief
Interaction between belief and outcome during foreshadow-first trials.
“ Attempted Harm” produces significantly greater activation than any other condition.
Integration Results (Belief First)
RTPJ: Significant main effect of outcome.
No interaction between belief and outcome.
Explaining Order Differences
Why might a different pattern of activation exist in belief-first as opposed to foreshadow-first trials?
Authors suggest that perhaps participants “double check” belief information more often when it is presented early.
dMPFC showed a significant main effect for negative over neutral beliefs for foreshadow first trials.
Suggests that dMPFC may be involved in processing belief valence.
RTPJ, and to a lesser extent PC and LTPJ, appear to be important for both encoding and integration of belief information.
During encoding, increased activation is observed when belief information (neg or neu) is present.
During integration, RTPJ appears to discriminate between different belief/outcome combinations.
dMPFC does not appear to play a specific role in belief encoding.
During integration, dMPFC is significantly more active for negative beliefs than for neutral beliefs.
Interpreting PFC results
Past studies have focused on the role of the MPFC in moral judgment.
These studies have been unable to distinguish whether MPFC is recruited for representing outcome valence or belief valence.
Results reported here suggest that MPFC is important for belief valence, and not outcome valence.
What do these results tell us about moral judgment?
Mental states are crucial in moral judgment, and have been substantially ignored until very recently (at least in the adult literature)
Studies that build off of the standard dilemma paradigm are problematically conflating belief and outcome information.
Future research on moral judgment should focus less on outcome information.
Although outcomes are clearly important when making a moral judgment, they are only a small piece of the information we use when making moral judgments.
What other factors might be involved (desires, prior behavior, causation, etc…)?
Penal Code may offer insights into the kinds of things that impact moral judgment.