Portal de Periódicos da CAPES

Dissociating the ability and propensity for empathy

2014; Elsevier BV; Volume: 18; Issue: 4 Linguagem: Inglês

10.1016/j.tics.2013.12.011

1879-307X

Christian Keysers, Valeria Gazzola,

Psychosomatic Disorders and Their Treatments

Neuroimaging suggests psychopaths have reduced vicarious activations when simply witnessing pain but less so when asked to empathize. This inspired us to distinguish the ability from the propensity to empathize. We argue that (i) this ability–propensity distinction is crucial to characterizing empathy in psychiatric disorders such as psychopathy and autism, (ii) that costly helping might be best predicted by the propensity for empathy, and (iii) suggest how social neuroscientists can start exploring this distinction. Neuroimaging suggests psychopaths have reduced vicarious activations when simply witnessing pain but less so when asked to empathize. This inspired us to distinguish the ability from the propensity to empathize. We argue that (i) this ability–propensity distinction is crucial to characterizing empathy in psychiatric disorders such as psychopathy and autism, (ii) that costly helping might be best predicted by the propensity for empathy, and (iii) suggest how social neuroscientists can start exploring this distinction. Many researchers agree that empathy is something like ‘feeling what we would feel in another's stead’. Instruments designed to measure empathy acknowledge multiple facets to empathy (e.g., cognitive vs emotional empathy [1Baron-Cohen S. Wheelwright S. The empathy quotient: an investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences.J. Autism Dev. Disord. 2004; 34: 163-175Crossref PubMed Scopus (2558) Google Scholar], fantasizing vs perspective taking vs personal distress vs empathic concern [2Davis M.H. A multidimensional approach to individual differences in empathy.JSAS Catalog Selected Documents Psychol. 1980; 10: 85Google Scholar], actions, emotions and sensations [3Keysers C. Gazzola V. Expanding the mirror: vicarious activity for actions, emotions, and sensations.Curr. Opin. Neurobiol. 2009; 19: 666-671Crossref PubMed Scopus (322) Google Scholar]; Box 1); however, individuals are characterized on each of these facets using a single variable on a high–low continuum. Inspired by a study on psychopathy [4Meffert H. et al.Reduced spontaneous but relatively normal deliberate vicarious representations in psychopathy.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (28) Google Scholar], here we argue, instead, for the dissociation of two variables – ability and propensity – for each facet of empathy (Box 2). We will discuss how this impacts on the science of helping behavior and social neuroscience.Box 1Empathy, vicarious activations and related conceptsVicarious activations: if being in a particular situation systematically activates a set of neurons, the reactivation of any of these neurons while witnessing another in the same situation is what we refer to as vicarious activation. Vicarious activations can relate to different facets: your actions, your sensations, your emotions, or even your cognitions.Operational definition of empathy: a witness W shows evidence of empathizing with object O (e.g., a human, animal or object like a robot) if W shows evidence (e.g., neural or behavioral) for the (vicarious) activation of actions, sensations, or emotions that he/she would activate in O's stead. The more the state of W during witnessing resembles the state W would have in O's stead (i.e., the more vicarious activation), the more W shows evidence of empathy. Empathy is quantified in relation to what W would activate in O's stead, not what O really activates – empathy can be high but inaccurate. We favor this umbrella definition including many related phenomena (e.g., motor mimicry, mirror neurons, emotional contagion, etc.) to facilitate an empirical study of empathy across species and methods. We therefore exclude the metacognitive requirement that W be aware that the vicarious activations represent not his but O's state because establishing whether a non-verbal animal or a neural activation is aware of who is being represented seems impossible. To use vicarious activations as a proxy for empathy does not mean that vicarious activations are necessary and sufficient for empathy. The relation between vicarious activations at the neural level and empathy at the experiential level remains to be clarified.Facets of empathy: empathy can be split into subdomains or facets. Some assume different facets a priori (e.g., cognitive vs emotional empathy [1Baron-Cohen S. Wheelwright S. The empathy quotient: an investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences.J. Autism Dev. Disord. 2004; 34: 163-175Crossref PubMed Scopus (2558) Google Scholar]); others extract them from a factor analysis of self-report questionnaires (e.g., fantasising vs perspective taking vs personal distress vs empathic concern [2Davis M.H. A multidimensional approach to individual differences in empathy.JSAS Catalog Selected Documents Psychol. 1980; 10: 85Google Scholar]), others still because neuroimaging reveals that different facets (e.g., actions, emotions, sensations, cognitions [3Keysers C. Gazzola V. Expanding the mirror: vicarious activity for actions, emotions, and sensations.Curr. Opin. Neurobiol. 2009; 19: 666-671Crossref PubMed Scopus (322) Google Scholar]) engage separable neural systems.Situation: when W witnesses a specific state of O, we use the term ‘situation’ broadly to cover the entire situation that triggers empathy (including O), to cover factors that influence how and how much W empathizes with O by altering for instance the goals of W (e.g., instructions), W's relation to O (e.g., information about O's race, fairness, gender, ingroup/outgroup), the cost of empathy or helping, and W's responsibility for O's state.Box 2The ability and propensity for empathy, and how to measure themAbility for empathy: the current potential of a witness for empathy under optimal situations. Ability is latent: it represents the upper bound of how strongly a person can engage a particular facet of empathy (Figure I). Because training or life experiences may alter ability over time, we refer to the current upper bound. Empathic ability does not require accuracy (i.e., W may feel what he would have felt in O's stead rather than what O actually felt). By analogy: the ability of a runner would be the highest speed she can run under optimal conditions.Propensity for empathy: the individual's tendency to empathize as a function of the situation. Because different situations will trigger empathy to different degrees in any given individual, the propensity of the person is a function of the situation. The propensity in a situation (ranging from 0 to 1) multiplied by ability determines empathy in that situation. In analogy, the propensity of a runner depends on situation: for some it is high in a race and low on a vacation beach, for others, high in both.Encouraging empathy to probe ability: because ability is the upper bound for that facet of empathy, ability cannot be directly measured: we cannot test all situations. However, placing the participant in a situation that strongly encourages empathy provides a proxy. Experimenters should use instructions (e.g., ‘Try to feel with the actor in the movie’) and objects of empathy (e.g., a romantic partner, a familiar person, a person of the witness ingroup) that encourage empathy. Confederate designs in which the participant and a confederate draw a straw to decide who will be the witness can help maximize ingroup feeling.Leaving things open to probe spontaneous propensity: how likely are you to experience empathy when nothing en- or discourages you from doing so? This spontaneous propensity is best measured using minimal instructions, such as ‘please look at the movies carefully’, and relatively neutral actors as stimuli.Discourage empathy to probe automatic propensity: morally the key question is, will you empathize even if doing so is bad for you? To probe this automatic propensity for empathy, one could discourage individuals from empathizing (‘try to be cold and detached’), distract them from empathizing (a demanding task that diverts attention from the emotions of others), or use situations in which empathy would interfere with goals (e.g., provide financial incentives for participants to take decisions that harm others). Vicarious activations: if being in a particular situation systematically activates a set of neurons, the reactivation of any of these neurons while witnessing another in the same situation is what we refer to as vicarious activation. Vicarious activations can relate to different facets: your actions, your sensations, your emotions, or even your cognitions. Operational definition of empathy: a witness W shows evidence of empathizing with object O (e.g., a human, animal or object like a robot) if W shows evidence (e.g., neural or behavioral) for the (vicarious) activation of actions, sensations, or emotions that he/she would activate in O's stead. The more the state of W during witnessing resembles the state W would have in O's stead (i.e., the more vicarious activation), the more W shows evidence of empathy. Empathy is quantified in relation to what W would activate in O's stead, not what O really activates – empathy can be high but inaccurate. We favor this umbrella definition including many related phenomena (e.g., motor mimicry, mirror neurons, emotional contagion, etc.) to facilitate an empirical study of empathy across species and methods. We therefore exclude the metacognitive requirement that W be aware that the vicarious activations represent not his but O's state because establishing whether a non-verbal animal or a neural activation is aware of who is being represented seems impossible. To use vicarious activations as a proxy for empathy does not mean that vicarious activations are necessary and sufficient for empathy. The relation between vicarious activations at the neural level and empathy at the experiential level remains to be clarified. Facets of empathy: empathy can be split into subdomains or facets. Some assume different facets a priori (e.g., cognitive vs emotional empathy [1Baron-Cohen S. Wheelwright S. The empathy quotient: an investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences.J. Autism Dev. Disord. 2004; 34: 163-175Crossref PubMed Scopus (2558) Google Scholar]); others extract them from a factor analysis of self-report questionnaires (e.g., fantasising vs perspective taking vs personal distress vs empathic concern [2Davis M.H. A multidimensional approach to individual differences in empathy.JSAS Catalog Selected Documents Psychol. 1980; 10: 85Google Scholar]), others still because neuroimaging reveals that different facets (e.g., actions, emotions, sensations, cognitions [3Keysers C. Gazzola V. Expanding the mirror: vicarious activity for actions, emotions, and sensations.Curr. Opin. Neurobiol. 2009; 19: 666-671Crossref PubMed Scopus (322) Google Scholar]) engage separable neural systems. Situation: when W witnesses a specific state of O, we use the term ‘situation’ broadly to cover the entire situation that triggers empathy (including O), to cover factors that influence how and how much W empathizes with O by altering for instance the goals of W (e.g., instructions), W's relation to O (e.g., information about O's race, fairness, gender, ingroup/outgroup), the cost of empathy or helping, and W's responsibility for O's state. Ability for empathy: the current potential of a witness for empathy under optimal situations. Ability is latent: it represents the upper bound of how strongly a person can engage a particular facet of empathy (Figure I). Because training or life experiences may alter ability over time, we refer to the current upper bound. Empathic ability does not require accuracy (i.e., W may feel what he would have felt in O's stead rather than what O actually felt). By analogy: the ability of a runner would be the highest speed she can run under optimal conditions. Propensity for empathy: the individual's tendency to empathize as a function of the situation. Because different situations will trigger empathy to different degrees in any given individual, the propensity of the person is a function of the situation. The propensity in a situation (ranging from 0 to 1) multiplied by ability determines empathy in that situation. In analogy, the propensity of a runner depends on situation: for some it is high in a race and low on a vacation beach, for others, high in both. Encouraging empathy to probe ability: because ability is the upper bound for that facet of empathy, ability cannot be directly measured: we cannot test all situations. However, placing the participant in a situation that strongly encourages empathy provides a proxy. Experimenters should use instructions (e.g., ‘Try to feel with the actor in the movie’) and objects of empathy (e.g., a romantic partner, a familiar person, a person of the witness ingroup) that encourage empathy. Confederate designs in which the participant and a confederate draw a straw to decide who will be the witness can help maximize ingroup feeling. Leaving things open to probe spontaneous propensity: how likely are you to experience empathy when nothing en- or discourages you from doing so? This spontaneous propensity is best measured using minimal instructions, such as ‘please look at the movies carefully’, and relatively neutral actors as stimuli. Discourage empathy to probe automatic propensity: morally the key question is, will you empathize even if doing so is bad for you? To probe this automatic propensity for empathy, one could discourage individuals from empathizing (‘try to be cold and detached’), distract them from empathizing (a demanding task that diverts attention from the emotions of others), or use situations in which empathy would interfere with goals (e.g., provide financial incentives for participants to take decisions that harm others). A hallmark characteristic of adults with psychopathy (PCL-R) and youths with conduct disorders of the ‘limited prosocial’ subtype (DSM-V) is reduced empathy. But are they unable to empathize, or simply less likely to empathize in certain situations? Psychopathic offenders [4Meffert H. et al.Reduced spontaneous but relatively normal deliberate vicarious representations in psychopathy.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (28) Google Scholar] were scanned while viewing people experiencing pain. They were first told to watch the movies carefully, then to deliberately feel what the individuals in the movies felt. Psychopaths showed reduced vicarious activations (Box 1) under instructions simply to observe, but these differences largely disappeared under instructions to empathize [4Meffert H. et al.Reduced spontaneous but relatively normal deliberate vicarious representations in psychopathy.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (28) Google Scholar]. Two groups can thus have a similar ability for vicarious activations, as revealed by situations encouraging empathy (‘try to feel with the victim’), but a different propensity to engage this ability spontaneously in situations not encouraging empathy (‘just watch the movies’). Psychopathic criminals can be charming and attuned while seducing a victim, thereby suggesting empathy, and later callous while raping a victim, thereby suggesting impaired empathy. If vicarious activations underpin empathy, the ability–propensity distinction derived from vicarious activations would translate to empathy, and allow for this complexity: any person with a high ability but reduced propensity for empathy could choose to use his ability for seduction but not for raping because empathy would facilitate the former but hinder the latter. To characterize empathy accurately in psychiatric groups it may thus be necessary to measure empathy in multiple representative situations to assess their ability, and to determine where along the propensity curve their empathy might be abnormal, and where it is preserved (Figure I in Box 2). This ability–propensity distinction may also shed light on autism [5Gillespie S.M. et al.Spontaneous versus deliberate vicarious representations: different routes to empathy in psychopathy and autism.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (221) Google Scholar]. Facial mimicry is abnormal when autistic individuals watch facial expressions, but not when deliberately imitating them, and mu wave suppression is reduced when they watch unfamiliar but not familiar others act [5Gillespie S.M. et al.Spontaneous versus deliberate vicarious representations: different routes to empathy in psychopathy and autism.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (221) Google Scholar]. Autism might therefore also be better described as an abnormal propensity rather than ability for empathy. The distinction we propose builds on similar distinctions in other domains. The International Classification of Functioning, Disability, and Health urges clinicians to distinguish motor capacity – what someone can do in an optimal lab situation – from motor performance – what a person actually does in the home environment. Similarly, psychometrics distinguish between competence and performance. For instance, a toddler could fail a false-belief task either because she lacks a mature theory-of-mind (low competence) or because the task requires other cognitive functions (e.g., working memory) that inadvertently limit performance. That changing instructions can boost vicarious activations reveals that empathy is not entirely automatic. Some of the processes behind the ability seem to depend on attention and/or motivation. Spontaneously, psychopaths show reduced amygdala activation and fear-conditioning, but these responses normalize if their attention is focused towards the emotional dimension of the stimuli – suggesting that abnormal attention might be crucial to psychopathy [6Larson C.L. et al.The interplay of attention and emotion: top-down attention modulates amygdala activation in psychopathy.Cogn. Affect Behav. Neurosci. 2013; 13: 757-770Crossref PubMed Scopus (98) Google Scholar]. This account begs the hypothesis that, more generally, the degree to which attention is automatically captured by social stimuli should predict propensity for empathy. Individuals seek an optimal level of connectedness, and this motivates those less connected to empathize more, and those more connected to empathize less with others (J. Zaki, submitted review). Helping others is costly, and therefore creates a motivation against empathizing; individuals therefore downregulate their empathy in costly helping situations by avoiding empathy-inducing stimuli (J. Zaki, submitted review). Empathizing with someone increases that person's trust, thereby facilitating access to that person's resources. These are only three of the many factors that could motivate people away from or towards empathy. Individual differences in attention and motivation could thus turn the knob of empathy up and down, creating individual differences in how strongly propensity and ability dissociate. Patients may then be unusual minima in the distribution of automatic social attention or motivation. Do autistic individuals empathize with fewer individuals [5Gillespie S.M. et al.Spontaneous versus deliberate vicarious representations: different routes to empathy in psychopathy and autism.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (221) Google Scholar] because their motivation for connectedness is restricted to fewer individuals? Experiments that systematically alter attention and motivation could explore these factors. Evolutionarily, an ability to sense the inner states of others is always adaptive: it improves the capacity to predict and adapt to the behavior of others. Always exercising this ability can however be costly: by motivating us to help others, it depletes our resources and render the capacity to downregulate empathy adaptive. The niche of a species should determine this downregulation. A competitive niche should select downregulation; a purely cooperative niche not. The mixed niche so typical of humans would select for a mix of individuals with high and individuals with low automatic propensity. Since David Hume and Adam Smith, helping behavior is thought to depend on empathy. However, more empathic individuals have only been found to be moderately more likely to help others [7Hoffman M. Empathy and Moral development: The Implications for Caring and Justice. Cambridge University Press, 2000Crossref Google Scholar] – but why? Because helping is costly, the cost increases motivation against empathy, and we would thus argue that costly helping of others should depend on the propensity to empathize ‘automatically’ (Box 2) – that part of empathy that occurs without motivation or attention – not on the ability. The association between empathy and costly helping might have been diluted in past research because empathy was measured without distinguishing between ability and automatic propensity. For example, self-report instruments do not always separate these two aspects: ‘When I watch a good movie, I can very easily put myself in the place of a leading character’ [2Davis M.H. A multidimensional approach to individual differences in empathy.JSAS Catalog Selected Documents Psychol. 1980; 10: 85Google Scholar] or ‘I can tune in to how someone else feels rapidly and intuitively’ [1Baron-Cohen S. Wheelwright S. The empathy quotient: an investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences.J. Autism Dev. Disord. 2004; 34: 163-175Crossref PubMed Scopus (2558) Google Scholar]: both mix an ability element (‘can’) and a propensity element (‘easily’, ‘rapidly’, ‘intuitively’). Social desirability might lead individuals to overestimate or report their automatic propensity for empathy. Designing instruments that separately assess ability and propensity, and where participants are placed in relevant situations, could test our prediction that the automatic empathic propensity would be a better predictor of costly helping than the empathic ability. Social neuroscience can examine vicarious activations as a proxy of empathy and explore whether and how a capacity–propensity distinction maps onto the brain. Such a mechanistic neural understanding could guide the development of interventions to up- or downregulate the ability or propensity for various facets of empathy. It would be vain to attempt simply to map the ability for vicarious activations on some brain regions and the propensity on others: if ability is an upper bound, it requires the synergy of all the regions involved in vicarious activations, including those involved in boosting empathy in the most empathy encouraging situations. Several techniques might however help distinguish two neural systems that relate to our distinction: a ‘core’ circuit for vicarious activations, that is always necessary for a given facet of empathy, and multiple ‘modulatory’ circuits that drive and modulate the core in specific situations. The propensity for empathy is then approximated by the profile of how strongly the modulatory circuits drive the core circuit as a function of situation; the ability, by the strongest activation of the core circuit via these modulatory circuits. The core circuit should be common to all situations triggering a facet of empathy, whereas modulating circuits should be situation-specific. Meta-analyses of emotional empathy show the anterior insula and the rostral cingulate cortex are activated consistently across very different paradigms triggering emotional empathy, revealing a ‘core circuit’ for emotional empathy. Core circuits also exist for empathy for sensations and actions (Figure 1A) . Other brain regions come online preferentially when emotional empathy is triggered by abstract cues or by explicit images (Figure 1B, blue and red, respectively) and represent ‘modulatory circuits’ for emotional empathy [8Lamm C. et al.Meta-analytic evidence for common and distinct neural networks associated with directly experienced pain and empathy for pain.Neuroimage. 2011; 54: 2492-2502Crossref PubMed Scopus (1402) Google Scholar]. Patients with lesions in the insula seem unable to attribute disgust to others for a wide range of stimuli, linking such a structure to the core circuit for disgust empathy [9Calder A.J. et al.Impaired recognition and experience of disgust following brain injury.Nat. Neurosci. 2000; 3: 1077-1078Crossref PubMed Scopus (650) Google Scholar]. Lesions in the amygdala impair spontaneous fear attribution, but if gaze is directed to the eye regions, the impairment disappears, linking such structures to modulatory circuits for empathy [10Adolphs R. et al.A mechanism for impaired fear recognition after amygdala damage.Nature. 2005; 433: 68-72Crossref PubMed Scopus (1018) Google Scholar]. Interestingly, amygdala responses are also often abnormal in psychopathy unless attention is directed to emotions [4Meffert H. et al.Reduced spontaneous but relatively normal deliberate vicarious representations in psychopathy.Brain. 2013; 136: 2550-2562Crossref PubMed Scopus (28) Google Scholar, 6Larson C.L. et al.The interplay of attention and emotion: top-down attention modulates amygdala activation in psychopathy.Cogn. Affect Behav. Neurosci. 2013; 13: 757-770Crossref PubMed Scopus (98) Google Scholar]). Using individual differences in propensity or diagnosis as regressors for resting-state or diffusion-weighted images could help identify connectivity features that predispose individuals to a specific propensity profile [11Ly M. et al.Cortical thinning in psychopathy.Am. J. Psychiatry. 2012; 169: 743-749Crossref PubMed Scopus (118) Google Scholar]. Research on attention and reappraisal has outlined circuits that allow individuals to process preferentially stimuli that suit their current goals and motivations [12Corbetta M. et al.The reorienting system of the human brain: from environment to theory of mind.Neuron. 2008; 58: 306-324Abstract Full Text Full Text PDF PubMed Scopus (2760) Google Scholar] and to regulate their own emotions in reaction to social stimuli [13Buhle J.T. et al.Cognitive reappraisal of emotion: a meta-analysis of human neuroimaging studies.Cereb. Cortex. 2013; https://doi.org/10.1093/cercor/bht154Crossref PubMed Scopus (1155) Google Scholar] (Figure 1C,D), revealing powerful modulatory circuits for empathy. Individual differences in empathy are somewhat stable across the lifespan [14Knafo A. et al.The developmental origins of a disposition toward empathy: genetic and environmental contributions.Emotion. 2008; 8: 737-752Crossref PubMed Scopus (363) Google Scholar], but what is the relative stability of our propensity and capacity for empathy? Which factors alter our propensity and which our ability for empathy? Specific experiences (e.g., piano lessons) can create new motor programs and thus new empathic abilities. Others (e.g., meditation; www.ReSource-Project.org) can encourage us to use our empathic abilities more habitually, thereby altering our propensity. Understanding how to trigger changes in our capacity and propensity, how stable such changes can be, and what neural circuits they rely on is an important research agenda for understanding individual differences and designing therapies that target specific deficits in patients. Deep analysis of how neurons contribute to the ability and propensity for vicarious activations, and how these shape the ability and propensity for empathy, will not be easy. However, replacing a univariate notion of trait empathy for each facet with a more multivariate landscape of capacities and propensities shaped by attentional and motivational factors will be necessary to capture the complexity of psychiatric disorders, to help these patients, and to understand the striking differences in helping across individuals and situations. C.K. is supported by a European research Council (ERC) grant ‘VicariousBrain’ (312511) from the European Commission, and V.G. by Veni grant 451-09-006 from the Netherlands Organisation for Scientific Research (NWO). We thank Dan Batson, Harma Meffert, Claus Lamm, and Michael Spezio for comments that helped shape this paper. We thank Claus Lamm for sending us the maps used to generate Figure 1B, and Jason Buhle, Jennifer Silvers, and Kevin Ochsner for sending us those to generate Figure 1D.