Investigating the Multidimensionality of State Empathy: Evidence from Behaviour and Event-Related Potentials

Abstract

Empathy is a critical facet of social cognition that is generally understood to refer to a set of related affective and cognitive constructs that shape how we respond to others’ emotional states. Many researchers, particularly those using event-related potential (ERP) methods to study empathy for physical pain, have claimed that conceptual distinctions between affective and cognitive empathy constructs are reflected in the neural time-course of empathy. These studies rarely record brain activity evoked during different tasks that tap into different empathy constructs; such designs cannot draw firm conclusions about the neural correlates of different types of empathy. Across three studies I manipulated empathy task type to determine whether state-level empathy ratings and neural dynamics during empathy for pain and fear were modulated by the specific type of empathy being tapped.

In Chapter 2, I developed a contextual priming design in which participants viewed images of body parts experiencing painful and neutral stimulation followed by faces making congruent pained and neutral expressions. I tested whether participants’ state-level empathy ratings varied across three tasks reflecting affective sharing (AS), empathic concern (EC), and affective perspective-taking (aPT), along with a control gender discrimination (GD) task. I also tested whether state empathy ratings were predicted by two different trait measures of empathy. For the first time, I demonstrated that there was a behavioural distinction between affective (AS) and cognitive (EC, aPT) empathy constructs at the state level. However, trait scales generally did not predict state empathy ratings, even when both measures supposedly tapped into the same construct.

In Chapter 3, I tested whether affective empathy is instantiated by modulation of early frontocentral components (N1, N2) while cognitive empathy is indexed by later centroparietal components (P3, LPP) as is often claimed. I used the same experimental paradigm as Chapter 2, time-locking ERPs to both stimulus types, and incorporated a large sample size, a gaze-contingent design, and robust, data-driven mass univariate statistics. Empathy was characterized by a widespread effect of affective state from 200-650ms after onset for body part stimuli, regardless of the empathy construct being tapped. This effect was absent for the control (GD) task. Face stimuli elicited earlier effects of pain (150ms) in general; pain effects for faces were also weaker and more temporally restricted during the aPT task relative to the AS and EC tasks. These findings are likely due to the influence of affective priming and task difficulty, suggesting that attentional effects may play a greater role in modulating the neural timing of empathy than the specific empathy construct being tapped.

In Chapter 4, I used the same paradigm but replaced the pained expressions with fearful expressions to determine whether the results of Chapter 3 would generalize beyond empathy for physical pain. ERPs evoked by body parts largely replicated the results of Chapter 3 as all empathy tasks elicited a similar pain-neutral difference from 200-650ms. Fearful faces produced similar findings compared to pain faces, with an effect of fear beginning around 100-150ms from onset; the effect of affective state was again much weaker during the aPT task specifically. However, in contrast to Chapter 3 there were also significant, weak effects of affective state during the control task for both stimulus types. These results indicate that in addition to task demands and priming, the neural dynamics of empathy may also be influenced by the specific affective context in which one empathizes.

In my final study (Chapter 5), I re-analyzed my neural data to determine whether ERPs evoked during empathy were truly associated with self-report measures of empathy as has been previously reported. Using a mass univariate approach, I found no meaningful correlations between trait empathy measures and ERPs evoked during Chapters 3 and 4, regardless of task type or affective context. I also employed trial-level regression models to assess the degree to which the effects of empathy, task type, and their interaction on ERPs were related to participants’ subsequent empathy ratings on each trial. Main effects of affective state were consistently related to empathy ratings, although the degree to which this varied across task types was inconsistent.

I discuss the findings of these studies in terms of their implications for theories of empathy, as well as how they can be interpreted in the context of the broader ERP literature on emotion, attention, and face processing. My results indicate that the neural time-course of empathy is not characterized by specific empathy constructs and can instead be understood in terms of an early (200-250ms; 100-150ms if primed) attentional orienting to salient affective stimuli followed by later (350-650ms) evaluative social cognition. The neural distinctions underlying different empathy constructs may instead be rooted in later neural processing or the activation of specific neural circuits that cannot be dissociated with ERP methods.