Vervoort et al. (2013) examined eye-tracking patterns among people observing facial expressions displaying low, medium, or high levels of pain paired with neutral expressions. Participants’ time to first fixation varied as a function of their self-reported levels of pain catastrophizing—the tendency to ruminate about, magnify the potential risks of, and feel helpless in the face of pain (Sullivan et al. 1995). Among participants low in catastrophizing, time to first fixation was more rapid to faces displaying pain and decreased with increasing intensity of pain expression. Among high catastrophizers, time to first fixation did not vary between neutral and pain expressions. This pattern suggests the operation, among people high in catastrophizing, of a (likely unconscious) preattentive process in which evidence of threat is rapidly processed and drives avoidance behavior. More generally, as catastrophizing is but one of a category of variables likely to prime a hypervigilant state (e.g., Yamada and Decety 2009), the framework includes this set of influences under the general rubric of negative affectivity. Future research is necessary to map and clarify the nature of such influences.

Detection is the process of determining that the behavior of the sufferer indicates pain. It is influenced by two variables: the magnitude of the evidence of pain and the observer’s sensitivity to that evidence. The earliest work to examine perceptual processes involved in third-person pain demonstrated that naïve observers were capable of discriminating variations in others’ pain intensity purely on the basis of facial expression (Prkachin and Craig 1985). Interestingly, the ability to discriminate differences in sufferers’ experience was independently related to the physical intensity of the stimuli used to evoke pain and the sufferers’ subjective reports of pain, suggesting that observers are sensitive to behavioral cues distinguishing sensory and affective dimensions. Although observers are sensitive in general to behavioral cues of pain, they appear to rely primarily on the more salient indicators and to be relatively insensitive to subtle signs (Prkachin et al. 1994). Observers also appear to be sensitive to features of the display that indicate active modulation on the part of the sufferer, such as faking or exaggeration (Prkachin 1992a, 2005).

Sensitivity to pain expression appears to be a rather robust ability that does not vary substantially across people. There is evidence, however, that sensitivity to pain in others can be affected by what the proposed framework refers to as intersubjective traits and processes—personal characteristics and dynamic variables that influence how one registers the experiences of others. For example, groups known for their social impairments have been found to differ in their sensitivity to pain expression. Martins et al. (2011) compared schizophrenics with healthy controls in their ability to detect facial expressions of pain. Schizophrenic patients displayed consistent deficits in the detection of facial expressions of pain across a spectrum of pain intensity. Wojakiewicz et al. (2013) replicated this finding with a more homogeneous sample of patients diagnosed with paranoid schizophrenia. This study showed, in addition, that sensitivity to the pain of others was correlated with empathic characteristics of fantasy (the tendency to imagine oneself in hypothetical situations) and empathic concern among normal individuals, but not in the schizophrenic group, suggesting a decoupling of features of affective information processing and self-awareness in the latter. Caes et al. (2012) found that psychopathic traits were associated with a diminished perceptual sensitivity to others’ pain. Lastly, pain judgments can be influenced by the sufferer’s characteristics. The foregoing findings have been interpreted to imply that empathy deficits may underlie diminished sensitivity to others’ pain. However, it is possible that other variables correlated with the characteristics under study are responsible for diminished sensitivity. Psychopathy involves multiple components (e.g., callousness) other than deficient empathy that could equally account for diminished sensitivity.

Another variable that has been shown to affect sensitivity to pain in others is likability (De Ruddere et al. 2011). Observers exposed to an evaluative conditioning procedure in which still photographs of pain patients were associated with adjectives reflecting positive (e.g., “faithful”), neutral (e.g., “conventional”), or negative (e.g., “arrogant”) traits rated the intensity of the pain displayed by the patients during range-of-motion exercises. When viewing video recordings showing high levels of pain expression, patients whose images had been associated with negative traits were rated to be in less pain than patients associated with neutral or positive traits, with the latter two conditions not differing. This effect was attributable exclusively to reduced sensitivity to higher levels of pain.

In their pain empathy model, Goubert et al. (2005) note that instead of displaying sympathy and concern to the person in pain (characterized as an “other-oriented” response), “self-oriented” observers may experience personal distress. There is reason to believe that such an orientation may also diminish sensitivity to the other’s pain by virtue of generating active avoidance of evidence, for example, by gaze aversion.

Having registered the occurrence of pain in another, the observer must ultimately make some kind of response, even if it is to do nothing but continue to observe. The process of response selection is dependent on elaborative and evaluative processes that relate evidence about the sufferer’s pain to other information the observer has available. In the evaluation/elaboration phase, the observer makes judgments about what the pain display means in relation to certain key dimensions. These include pain intensity, suffering, the credibility of the sufferer’s display, and personal characteristics of the sufferer or his or her circumstances, such as the degree to which they have been responsible for their misfortune. The degree to which the sufferer’s pain poses a personal threat to the observer is also evaluated.

The most intensively studied of these dimensions is pain intensity. Observers’ judgments about the pain of others are typically about how much pain the other is feeling or how unpleasant it is. In empirical studies, observers typically render judgments by using a quantitative scale. Observers have preferences for where they place their ratings on such scales. Some are conservative and tend to make lower ratings; some are liberal, distributing their ratings toward the higher end. These tendencies are called response biases. When observers make use of the same scales that sufferers have used to rate their own pain, it is possible to evaluate how closely third-person ratings match first-person ratings.

Prkachin and Mercer (1989) studied patients attending a physiotherapy clinic for treatment of injuries to their shoulders. As part of their assessment, the patients underwent a series of tests in which their affected shoulders were maneuvered through a range of motion. At the end of each test, the patients rated how much pain it caused. In a subsequent study, observers were shown video recordings of patients’ facial expressions and rated how much pain they thought the patient experienced, using the same scale the patients had used (Prkachin et al. 1994). The results indicated that although judges’ ratings tracked the patients’ in terms of the painfulness of different tests, they were significantly and substantially lower than those of the patients. This effect has been termed an “underestimation bias.” (It must be acknowledged that the term reflects the perspective of the sufferer and should not imply an ontological reality since it is really a reflection of the comparison of two subjective estimates.)

In a later study (Prkachin et al. 2001) a similar methodology was used to study groups of observers who differed with respect to the extent and nature of their experience with pain. One group consisted of people who had lived with a person who suffered from chronic pain, while another consisted of therapists whose practices involved treatment of people with pain conditions. A third group consisted of people with no significant experience with pain in others. All participants rated the video recordings of patients with shoulder pain, using the same scales the patients had used. Results indicated that all groups of observers displayed the underestimation bias. There were interesting differences in the rating patterns of the different groups, however. Relative to observers who had little experience with pain in others, the ratings of people who had lived with a pain sufferer approximated those of the patients more closely. By contrast, people with substantial clinical experience working with pain sufferers showed an enhanced underestimation bias. The finding of an enhanced underestimation bias among health professionals when judging others’ pain on the basis of pain expression is consistent with a broader literature indicating that various health professionals tend to underestimate the pain of others when relying on diverse sources of information (Prkachin et al. 2007; Solomon 2001) and with research showing that physicians give lower ratings of pain and fail to demonstrate electroencephalographic differentiation in frontal and centro-parietal regions when observing painful versus nonpainful stimulation to others (Decety et al. 2010). The implication is that something about the experience of health-care provision is responsible for a systematic alteration in processes for judging others’ pain. Importantly, however, the finding that people who have lived with a pain sufferer show a diminished bias equally implicates experiential influences on pain judgment processes, but suggests that experiential effects can operate in different directions. Of critical concern is determining the kinds of influences that can shape judgments in either direction.

The classical perception theory of adaptation level (Helson 1964) provides one basis for explaining differences in pain estimation judgments. Adaptation-level theory holds that the evaluation an observer makes about a stimulus is determined in part by the context in which the stimulus is presented. The same stimulus judged in the context of weaker stimuli is perceived to be stronger than when it is judged in the context of stronger stimuli. In a study that bears on the adaptation-level concept, Prkachin et al. (2004) presented participants with very brief videos of the facial expressions of patients displaying no pain or moderate pain. Participants indicated whether they thought patients were displaying pain or not. Four experimental conditions manipulated the degree of exposure that each participant had to displays of strong pain. Exposure to 1, 5, or 10 displays of strong pain resulted in a reduced likelihood of judging people to be in pain, relative to no exposure, with the biasing effect evident at the lowest level of exposure and increasing with greater exposure. In short, experience with displays of strong pain led to a diminished likelihood of imputing pain to others. Prkachin and Rocha (2010) replicated this effect. The parallel with clinical scenarios is evident. Health professionals who work with pain sufferers are exposed to frequent and high levels of pain expression. The cumulative effects of such exposure are likely to set professionals’ adaptation levels higher than those of people exposed to evidence of suffering to a lesser degree, making them more susceptible to underestimation.

Attitudinal factors reflecting observers’ attributions about the motivations of sufferers have also been shown to influence third-person pain judgment biases. As emphasized by Craig et al. (2010), although they occupy an intermediate position on the automatic-deliberate continuum, nonverbal expressions of pain are under some degree of conscious control, and there is evidence that the topography of deliberately modulated expression differs in subtle ways from that of spontaneous expression (Craig et al. 1991; Galin and Thorn 1993; Hill and Craig 2002; Prkachin 1992a, 2005). Williams (2002) emphasized that evolution should have prepared observers to be sensitive to this possibility in the form of a “cheater detection” algorithm. Kappesser et al. (2006) showed videos of shoulder pain patients displaying various levels of pain to health-care providers (primarily emergency room physicians and nurses) and had them rate the patients’ pain using the same scales the patients did. Observers participated in one of three conditions. In one, they were told the actual rating the patient had given. A second group was also informed that the patients had been seeking opioid medication for pain relief. The control group was given no information. Results showed that providing information about the participants’ own ratings partially diminished the usual underestimation effect; however, the addition of information indicating that patients were opioid seeking effectively eliminated the benefit attributable to learning of the patient’s own rating. Furthermore, observers’ estimates of the base rate of exaggeration of pain in patient populations were associated with enhanced underestimation, regardless of the condition to which they were assigned.

The judgments rendered in the evaluation/elaboration phase will, in principle, bias the observer toward or away from certain actions. The more intense the observer judges the person’s pain to be, the more likely he or she is to be helpful. The observer who judges the sufferer to lack credibility may ignore or even engage in aggressive behavior toward him or her. In our framework, a series of response options, ranging from aggressive behavior on the one hand to engaging in behavior that is personally costly, is shown in a continuum. Unfortunately, detailed knowledge about the response options that people select is not available largely because of the difficulties entailed in studying them systematically. Nevertheless, some work has begun to address this component.

Hein et al. (2011) measured self-reported affect and skin conductance responses while participants were exposed to painful electric shocks. A confederate who was present was portrayed as receiving the same shocks as the participant. In a second session, the participant was given the choice of helping the other person by taking their shocks, watching but not helping the other or watching a video. Skin conductance responses while watching the other apparently receiving shocks were positively correlated with participants’ ratings of the other’s pain and predicted a greater likelihood of choosing the personally costly option of accepting pain for the other. Also, the more similar the participant’s skin conductance responses while observing pain to the other person were to the participant’s own responses, the more likely was costly helping.

Other research, while not directly measuring overt response choices, have has made use of proxy measures. Because most of this work has addressed issues of clinical relevance, it is reviewed below.

Ultimately, we must be interested in how the process of third-person pain plays out for the sufferer. This will be a matter of the behavioral choices the observer makes in the face of evidence of suffering. Of particular interest are the long-term effects of observers’ responses, such as those that would occur in the common scenario of a chronic, intractable pain condition. Although this is perhaps the most important issue of all, it is here where we have the least information and where research is needed the most. How, for example, is a tendency to underestimate another’s pain likely to affect the sufferer in the short and long term?

The behavior of the observer, or observers, over time, will have implications for the sufferer’s affective responses, their behavior and, in principle, their pain, and issues related to it. Anecdotal evidence, such as qualitative and survey studies, suggests that dismissive behaviors on the part of the observer might be predicted to follow from an extreme underestimation bias or judgments of diminished credibility produce negative emotional reactions (Herbette and Rime 2004; Morley et al. 2000). Anger, depression, and a sense of injustice have all been implicated as consequences of persistent pain, and plausible arguments can be constructed around how the responses of others to one’s pain would contribute to them.

The sufferer’s behavior might be affected positively or negatively by the behavior of others. From the perspective of the influential behavioral approach to pain (Fordyce 1976), withholding of positively reinforcing behavior on the part of an observer might be expected to diminish the sufferer’s pain-related behavior. On the other hand, theories of treatment adherence would likely predict that any behavior on the part of an observer that diminishes satisfaction with care would be associated with noncompliance on the part of the sufferer.

Each of these speculative effects on the affect and behavior of the pain sufferer would be likely to affect the long-term outcome of the condition. This is an area in which there is virtually no empirical evidence to inform discussion. Given the importance of the issue and the fact that there are practical ways of addressing it, it should be a priority for future research.

The predominant realm in which third-person pain makes a difference is health care. Identification of pain in others is one of the most important responsibilities of health-care providers. For individuals who are not capable of communicating effectively verbally, such as infants and people with dementia, nonverbal expression is the only overt means by which their suffering can be conveyed (Craig 2006; Hadjistavropoulos et al. 2011a). Hence, the basic attentional, perceptual, and cognitive processes that give rise to third-person pain—sensitivity to the behavioral display and judgment criteria concerning how much evidence to identify pain is necessary—are a first line in the alleviation of suffering.

A different set of relevant circumstances involves clinical scenarios characterized by ambiguity. These include cases in which the underlying source of the pain is unclear, where situational, personality, or behavioral factors introduce the possibility that additional variables may be complicating the evidence at hand, or where the course of action may involve the risk of increasing pain or inflicting damage. Scenarios like this are common with many subacute or chronic conditions, such as low back pain.

The critical issues that arise in these circumstances are the response that is evoked in the health-care provider and the behaviors that he or she is likely to engage in. Behavioral evidence (Prkachin et al. 2007) suggests that the natural inclination to underestimate the pain of others is enhanced among health-care providers and that this exaggerated bias may be a direct consequence of high-density exposure to the suffering of others (Prkachin et al. 2004; Prkachin and Rocha 2010). Cheng et al. (2007) showed that, relative to inexperienced controls, acupuncturists with considerable experience with inflicting pain on others displayed virtually no hemodynamic signal change in the insular cortex, cingulate cortex, and other relevant regions when observing simulations of painful needle insertions. The authors suggested that these differences reflect an experientially based adaptation that inhibits empathic responding allowing experts to engage in practice without becoming emotionally overwhelmed.