Contribution of self- and other-regarding motives to (dis)honesty

Abstract

Why would people tell the truth when there is an obvious gain in lying and no risk of being caught? Previous work suggests the involvement of two motives, self-interest and regard for others. However, it remains unknown if these motives are related or distinctly contribute to (dis)honesty, and what are the neural instantiations of these motives. Using a modified Message Game task, in which a Sender sends a dishonest (yet profitable) or honest (less profitable) message to a Receiver, we found that these two motives contributed to dishonesty independently. Furthermore, the two motives involve distinct brain networks: the LPFC tracked potential value to self, whereas the rTPJ tracked potential losses to other, and individual differences in motives modulated these neural responses. Finally, activity in the vmPFC represented a balance of the two motives unique to each participant. Taken together, our results suggest that (dis)honest decisions incorporate at least two separate cognitive and neural processes-valuation of potential profits to self and valuation of potential harm to others.

Figure 1

Trial timeline. On each trial, the participant (a “Sender”) chose which message to send to the Receiver, out of four options. The text at the top of the screen is the message that would be sent on this trial to the Receiver. Four options are revealed, each one consisting of some amount of money for the Sender (“self”) and some for the Receiver (“other”). One option was always truthful (the one more beneficial for the Receiver; #2 in the example) and one deceptive (#4). Payoffs to both players and locations varied between trials. The Sender had 6 s to indicate her choice, after which the chosen option was highlighted and stayed on the screen for the remainder of the trial.

Figure 2

Behavioural results (n = 28). (a) Reaction times correlate with overall dishonesty. The average difference between Lie reaction times and Truth reaction times are on the y-axis, and percentage of dishonest trials on the x-axis. Each circle represents a subject. (b) Participants are arranged from most honest (far left) to most dishonest (far right). Bars represent each participant’s regression coefficients, reflecting how much self-interest and regard for others contributed to their probability to lie. Greyed-out bars indicate non-significant coefficients.

Figure 3

Neural sensitivity to values (n = 27). (a) Amount of money for the Sender in the chosen option vs. the unchosen option. The activation map was masked using value-related ROIs taken from a meta-analysis (Bartra et al., 2013). Map at q(FDR) = 0.05. (b) Voxels sensitive to Sender’s potential profits from dishonesty (value for self; ΔV_self) (top). For visualization purposes, the map is thresholded at p = 0.001, cluster-size corrected. The BOLD response in the LPFC is positively correlated with the behavioural measure of self-interest (bottom). (c) Voxels sensitive to Receiver’s potential losses from dishonesty (value for other; ΔV_other) (top). Map thresholded at p = 0.005, cluster-size corrected. The BOLD response in the rTPJ is positively correlated with the behavioural measure of regard for others.

Figure 4

Value to self vs. value to other, modulated by preferences (N = 27). Contrasting potential profit for Sender (value to self; ΔV_self) with potential losses to Receiver (value to other; ΔV_other), modulated by individual differences in the balance between regard for other and self-interest. The activation map was masked using value-related ROIs taken from a meta-analysis (Bartra et al., 2013).