Attentional biases are a selective attention towards or away from a stimulus, which is both specific and salient to an individual’s current environment and situation [1
] and can result in a variety of cognitive, behavioral and physiological responses. Excessive attentional biases towards pain have been hypothesized to contribute towards the promotion of pain-related anxiety, fear of pain-related activity, physical disability and exacerbations in the pain experience [2
]. Attentional biases towards pain-related stimuli have been proposed in theories of attention and pain [3
]. The pain-specific models explain an individual’s attentional response to the presence of pain-related stimuli. Todd et al. [5
] proposed the threat interpretation model of attentional biases, suggesting that there is a relationship between an individual’s interpretation of threat, which then influences their attentional bias towards it. As threat interpretation increases, initial vigilance towards pain-related stimuli increases; the level of the threat then influences whether the individual is able to disengage from the threat or avoid the threat.
] have reported that pain participants have an attentional bias towards threat related information compared to controls, although with low effect sizes of 0.1–0.3 which may be due, in part, to the lack of consistency with the type of pain-related stimuli (e.g., words or pictures; sensory or affective pain). Within the most recent review [8
], the majority of studies used word-based stimuli or pain-related faces, rather than pictures associated with movement. Kourtzi and Kanwasher [9
] have identified that pictures which have implied physical movement lead to greater activation of the extended motor system, compared to images without implied movement. Therefore, it could be suggested that the increased motor activity associated with movement-related images might represent a marker of a more valid cognitive response to movement-related emotional stimuli.
Attentional biases have been assessed, traditionally, using the dot probe paradigm [10
]. This technique captures a momentary cross-section of attention at the stimulus offset. Therefore, the method does not indicate duration of attention or attentional exertion to stimuli [11
]. The use of eye-tracking equipment, however, can address some of these limitations and allows researchers to record location and duration of gaze fixations. In addition, eye-tracking also provides the opportunity to measure pupil diameter as an index of attentional effort [12
]. Eye gaze markers can, therefore, provide researchers with a range of additional metrics to improve the understanding of attentional biases in chronic pain patients.
Several studies have utilised the dot probe method in conjunction with eye tracking and identified a bias towards pain-related information. Unfortunately, however, these studies were all conducted with pain-free undergraduate students, using pain-related words [13
] or faces [15
]. While these studies do provide some insight into the mechanisms associated with attention to pain-related information, further research is needed to be conducted in patients suffering from chronic pain. To our knowledge, only three studies have used eye movements to examine attentional biases to pictorial stimuli, and these were with chronic headache populations [18
] and mixed chronic pain groups [20
]. Both Liossi et al. [18
] and Schoth et al. [19
] used a visual search task to assess the attentional biases of chronic headache patients. Participants were shown four facial expressions (pain, angry, happy and neutral). Both studies [18
] identified that the pain group demonstrated a significantly higher proportion of initial fixations on the pain face, compared to the other facial expressions. Furthermore, Liossi et al. [18
] found that patients had an initial shift in their attention towards pain stimuli, but then maintained their gaze on happy” images. There was no evidence that pain patients maintained their gaze on pain related images. These findings support the theoretical models of pain and attentional biases [3
] and propose that attentional biases may play an important role in the maintenance of chronic headache. Fashler and Katz [20
] investigated the attentional biases of undergraduate students who were experiencing a variety of chronic pain conditions (e.g., neck/back, migraine, ankle/knee, stomach, hip, arm, eye and jaw pain). Participants completed a dot probe task using injury related (e.g., needle being inserted into the skin, black eye, open wound, burned skin) and neutral images while their eye movements were recorded. Reaction time results revealed that chronic pain individuals responded faster to neutral stimuli in contrast to the injury related images. In contrast, the eye tracking data demonstrated that chronic pain individuals maintained attention towards injury related pictures. Supporting Todd et al.’s [5
] theory that as the interpretation of threat increases, vigilance towards the threat also increases. To date, however, no study has investigated the attentional biases of a chronic back pain patient population using both the dot probe paradigm and eye tracking approaches.
The aim of this study, therefore, was to provide a more in depth investigation of chronic back pain patients’ attention to pain-related images in comparison to non-symptomatic controls by recording eye movement behavior whilst participants also completed a dot probe task. We hypothesized that chronic back pain participants would have: (i) a significantly higher percentage of fixations to threatening stimuli compared to controls; (ii) a longer average fixation duration to threatening images; and (iii) exhibit a faster reaction time to threatening images in the dot probe task.
This study used an eye gaze protocol to consider the attentional biases of individuals with chronic back pain to threatening and neutral images using a modified dot probe paradigm. The eye gaze data highlight important new findings about the differences in attentional bias to threatening and neutral images between individuals with chronic pain and non-symptomatic controls that cannot be identified by using the standard dot probe paradigm.
Within-group analysis demonstrated that the chronic pain individuals had a significantly higher percentage of fixations, larger pupil diameter and longer average fixation duration on threatening compared to neutral images. There were no significant differences in eye gaze metrics to threatening or neutral images in non-symptomatic controls; the absence of attentional bias suggesting that the back pain specific images were not perceived as threatening for the non-symptomatic controls. Between-group analysis revealed that chronic pain participants also had a longer average fixation duration and larger pupil diameter to threatening images compared to the non-symptomatic controls. Chronic pain patients also had a faster initial fixation to the pain related image compared to the neutral image; the opposite pattern was found for the non-symptomatic controls. In contrast, and of concern to the validity of the standard dot probe procedure, there were no significant differences in attentional biases for the dot probe task between the chronic pain group and non-symptomatic controls. Taken together, the findings of this study suggest that eye gaze metrics may provide a more sensitive measure of attentional bias compared to the dot probe response time.
The majority of previous research investigating pain-related attentional biases using eye gaze has been within a non-symptomatic population and used word-based stimuli or pain-related faces [13
]. To our knowledge, only three studies have previously used back pain-specific images in a dot probe task [25
], but none have used eye-tracking markers concurrent with the dot probe test with pain-related physical activity movements. Consistent with the findings of Roelofs et al. [25
] the dot probe data indicated that there was no difference between chronic pain participants and non-symptomatic controls for attention to threatening images in congruent trials. Whereas, the eye gaze behavior in this study demonstrated that chronic pain participants attended to the pain-related images significantly faster, more often and for a significantly longer average duration than neutral images compared to non-symptomatic controls. The attentional bias to threatening stimuli in the pain group may have been due to the implied motion cues within the image. Kourtzi et al. [9
] found greater activation of the medial temporal/medial superior temporal cortex (MT/MST) when viewing photographs with implied movement and imaging studies have supported the role of these brain areas in the analysis of movement, but not object, recognition [28
]. Action understanding depends, in part, on prior knowledge about the movement’s goal and intention with predictions about an object’s (or body’s) future position being made from the motion implied in the static image [29
]. Participant’s memorial biases modulate the increase in attentional bias to threatening information and, therefore, they perceive implied painful motion in the image (e.g., rotation of the back, which causes them pain); the dynamic images accessing a more meaningful motor representation that is presented for analysis through the variety of eye gaze metrics employed in this study.
Eye tracking studies within healthy populations have identified vigilance towards pain-related words and faces [14
]. In line with previous research [18
], we identified that chronic pain patients attended to pain related images significantly faster than neutral images and in contrast to the profiles of non-symptomatic controls. Todd et al. [5
] have proposed the Threat Interpretation Model of pain that can be used to consider some of the study’s findings. Specifically, there is an initial vigilance towards the threat-related stimuli; as the threat continues, there is an avoidance of the threatening cues. Further, Fashler and Katz [20
] identified that both pain and control participants responded faster to neutral images in the dot probe task, whereas the eye movement data demonstrated that in the early phase of the trial, participants attended to the neutral images, then in the later phases, gaze duration was maintained on the injury-related images. These findings are in contrast to those presented here; there were no differences found for reaction time data and chronic pain patients only made faster first fixations to pain related images. Due to the short presentation time used in this study, we were unable to assess whether participants would maintain their gaze on the pain stimuli or, in line with Todd et al., [5
] the patients showed an active avoidance of the threatening cues. One possible explanation for the differences could be due to the specificity of the images. Fashler and Katz [20
] used images for which both individuals with, and without pain could perceive as threatening (e.g., an open wound, needle inserted into the skin etc.). In contrast, our study used images that were specific to back pain and were not deemed threatening for the control group. Todd et al. [5
] propose that initial vigilance occurs when participants interpret the stimuli as threatening. The specificity of the image, and using a homogenous pain group, may, therefore, have influenced the attentional bias to pain related stimuli. Furthermore, the differences between the studies could be due to the image presentation time and type of probe stimulus. In the current study, the images were presented for 500 ms and the dot probe was either.. or : and each probe could appear on either side. In contrast, Fashler and Katz [20
] presented the images for 2000 ms and participants were shown a single dot, the response required participants to press one of two buttons to indicate the side of the screen the dot was on. The significant differences in the methods between the studies may explain the differences in the reaction time data. In contrast to Fashler and Katz [20
], Yang et al. [30
] found that chronic pain participants had an early attentional bias towards catastrophe-based words, followed by avoidance of pain words. Due to the short stimuli presentation time within our study, we were only able to assess initial vigilance, consistent with previous research in pain populations [18
]. This bottom up process is considered to be automatic in anxious individuals. In contrast to other studies identifying a vigilance-avoidance pattern [18
], this study used images of physical activity (rather than words) associated with pain. Images with implied movement provide elements of action understanding. They may give more personal meaning, agency, ownership and motor response [31
] for the viewed activity for chronic back pain participants compared to words, pain faces or images of an individual experiencing pain (e.g., someone grimacing while they are completing an activity). Future studies should present pain-related images for a longer duration to identify whether chronic back pain participants maintain vigilance or attend then avoid threatening images. Using images that are specific to the pain condition may also provide researchers with the ability to alter biases to allow for top down processing of attending to goal directed information. This enhanced attention to goal related information may reduce avoidance of activity and disability levels. Although a greater understanding of attention in this context is needed, preliminary evidence has supported the therapeutic benefits of attentional bias modification in pain populations [32
In this study, chronic pain participants showed a significantly larger pupil diameter to threatening images compared to controls. As well as for light intensity, the pupil dilates under conditions of high attentional allocation and also in response to emotionally-congruent information [34
]. In this regard, it has been suggested that pupil dilation is a physiological response that can indicate brain mechanisms associated with the processing of emotional information [36
]. When viewing threatening, emotional images pupil dilation has also been found to be mediated by increased sympathetic activity (e.g., increased heart rate and skin conductance) [34
]. According to the biopsychosocial model of pain [37
], there is both a psychological and physiological response to pain. If individuals attend to their physiological reaction to emotional information, it increases their worry-based anxiety, based on schemas, causing them to associate the increased arousal with the pain and lead to active avoidance behaviors. The results from this study suggest that not only are individuals with chronic pain attending to pain-related images more than controls, but they are also allocating greater visual attention to them indicating that the stimuli having greater emotional congruence and meaning. This study suggests that investigating pupil diameter could be a useful addition to the study of chronic pain.
The eye gaze data from this study provides support for current models of attention and extends the current chronic pain literature. Models of attentional bias within chronic pain attribute slightly different roles to the process of attention. In general, they propose that individuals in pain are fearful of, and threatened by, pain [2
], which causes them to over attend to pain-related information. Pain is prioritized over other demands for attention [39
], which interferes with movement, leads to higher levels of anxiety, catastrophizing and ultimately exacerbation of levels of disability [40
]. Understanding the mechanisms of attention to information which is perceived as threatening, is essential to better understand approaches to effective intervention.
The longer duration of attention to pain-related images in this study may be a function of the schemas associated with the movements. During experimental debriefs, some of the chronic pain participants commented that the images they viewed reminded them of activities in which they would expect to experience pain. Beck [41
] proposed that maladaptive schemas cause individuals to have a preoccupation with threatening information and subsequently catastrophize due to a negative interpretive and attentional bias. Pincus and Morley [4
] proposed the schema enmeshment model of pain (SEMP) in an attempt to explain recall bias in chronic pain patients through the operation of schemas. Pain schemas contain sensory, intensity, spatial and temporal features of pain, while illness schemas contain information about the consequences of illness, and self-schemas contain information about the self. The chronic pain experience may have illness related schemas about the implications of self-future activity, which causes enmeshment with the pain sensory, self and illness schemas. The anxiety experienced by the pain participants with particular activities they perceive as pain threatening leads to different behavioral and cognitive reactions [1
]. The misinterpretation of pain stimuli leads to excessive fear of physical activity and avoidance of physical activity. This may be due to their negative cognitions and increased somatic anxiety about completing the activity, which leads to physical avoidance behavior [42
]. Current interventions in the UK tend to focus on reducing a patient’s disability through improved education about their pain and the opportunity to discuss problems with particular activities [43
]. Understanding the cognitive and emotional mechanisms behind a patient’s initial activity avoidance behavior may allow for more specific interventions that modify the way chronic pain patients not only attend to activity-related information, but also the way they interpret the planned movement. The interaction of these two cognitive biases will affect the processing of information, lead to behavior change and, potentially, reduce the patient’s disability.
There were some limitations in the study. First, the PHODA images used were not assessed for their affective content (valence and arousal) or whether the images were personally meaningful to the individual. The images have, however, been used successfully in previous dot probe studies and to assess perceived harmfulness of daily activities [44
] suggesting they have good ecological validity. Although we did not ask chronic pain participants to rate the images directly, in follow up manipulation checks, pain participants reported that they could attribute the images to behaviors in their daily life and that the activities would be difficult and painful for them to complete at home. Similarly, verbal follow up checks with the control group demonstrated that the PHODA images represented a “neutral” image set and they showed no bias to either image type. Therefore, the biases predicted in the chronic pain group can be attributed to the pain-specific content of the image set. Future studies should assess the valence and arousal of the images in larger and more varied pain populations. Future research should also consider asking patients to rate their personal relevance to them; there may be different attentional biases towards images that are more personally-relevant, compared to those that are not (e.g., see Lang’s meaning propositions within Bioinformational Theory [46
Due to the recruitment strategy, there may be some selection bias. It has been suggested that pain patients who take part in research are often highly motivated, have more severe pain and respond better to treatment [47
]. This is a continuing issue for pain research. We were, however, interested in the patients’ cognitive response to threatening stimuli and did not implement an intervention. Future research may wish to consider comparing the health status of patients who volunteer, compared to those who decline to take part in a study.
Although the power analysis indicated there was sufficient power for this study, the sample size was low in comparison to other studies investigating attentional biases. Therefore, future research should aim to include a larger sample size. It should be noted that although the use of eye gaze is regarded as a more direct assessment of attention, it does not reflect overt attentional engagement. For example, visual attention can occur in the absence of eye movements [48
] and eye tracking technology does not measure peripheral vision, which can be used to complete the dot-probe task accurately [17
Despite these limitations, the present study provides further support for attentional biases towards pain-related images in chronic back pain participants. This finding is supported by additional data from eye gaze measurement techniques that provide a richer and more detailed analysis of the attentional biases in this population. Future studies should be conducted which investigate whether images that involve movement are more reliable at identifying attentional biases in chronic pain patients compared to pain-related words within a dot probe paradigm. Future research should also investigate the relationship between attention and interpretation of pain to provide an updated model to explain the mechanisms associated with patient responses to chronic pain. If the cognitive mechanisms of attention cause individuals with chronic pain to attend and dwell on painful stimuli, interventions that focus on modifying a patient’s attention to goal-related outcomes (e.g., attentional bias modification) may have important beneficial effects on future quality of life.