The influence of prior information and experience on perceptual
decision making in pain
Malvika Godara
University of Amsterdam (10895450)
Angelos-Miltiadis Krypotos
University of Amsterdam
Tom Beckers
University of Amsterdam & KU Leuven
Fear learning and prior information seem to influence the decision-making process regarding the intensity of a pain stimulus or stimuli. But a recent study by Wiech et al., (2014) demon-strated that perceptual decision-making in pain is biased towards expected sensation when participants expect a neutral stimulus to be followed by the administration of a low or high intensity shock. However, the influence of prior information on pain (the intensity of pain stimulus to be received) in the presence of differing experience has not been studied so far. In the current study, participants receive information about which of the two cues (circle or square) presented will be followed by a high or low intensity shock. The participants expe-rienced shocks which were either congruent with the prior information or incongruent with the information. The results indicate that perceptual decision-making is not biased by prior information concerning visual cues and intensity of pain. Further, diffusion model findings suggest that the process of reaching a decision regarding pain is impacted by a combination of prior information about pain and the actual experience of pain.
Introduction
Pain, in its physical sense, is important because it indi-cates presence of harm and the ability to remedy pain al-lows individuals to protect themselves. But in case of chronic pain, such as arthritis or multiple sclerosis, it can be disabling and affect the cognitive functioning of a person. Individu-als suffering from chronic pain tend to experience deficits in working memory, loss of attentional control, and di fficul-ties with information processing and problem solving (Hart, Wade & Martelli, 2003). In order to reduce people’s su ffer-ing from chronic pain and improve their quality of life, there is a need to understand the mechanism through which pain is perceived and how it affects individual cognitive function.
The process of pain perception (i.e., how a person identi-fies pain) has been found to be inferential (Apkarian, Hasmi & Baliki, 2011; De Ridder, Elgoyhen, Romo & Langguth 2011), suggesting that the perception of pain varies depend-ing upon the person and context within which pain is experi-enced. Several factors influence this process such as expecta-tions, emotions/emotional states, memory, and prior experi-ence (Zaman, Vlaeyen, van Oudenhove, Wiech & van Diest, 2015). These factors, specifically prior information, influ-ence the process of sensory-discrimination in pain (Wiech et al., 2014). Perceptual decision-making, which is the pro-cess of making decisions using the available sensory infor-mation (Wiech et al., 2014; Heekeren, Marrett, Bandettini & Ungerlelder, 2004), is altered by these factors (Wiech et
al 2014; Karim, Harris, Morley & Breakspear, 2012; Ahs, Miller, Gordon & Lundstrom 2013; Li, Howard, Parrish & Gottfried, 2008; Dunovan, Tremel & Wheeler, 2014). Role of Prior Information
Expectations are crucial to the process of pain percep-tion and tend to modulate perceptual decision-making. It has been found that expectations bias perceptual decision-making (Dunovan et al., 2014). To indicate, negative expec-tations can lead to perception of more severe pain and pos-itive expectations can cause a feeling of pain relief (Tracey, 2010). For example, an expectation of pain upon walking is likely to induce perception of greater pain upon walking, while an expectation of lesser pain upon using a stick to walk is likely to induce feeling of greater pain relief (see also placebo effect; Stewart-Williams & Podd, 2004). Ex-pectations of pain or pain relief can develop based upon prior information regarding pain, such as intensity or duration of pain sensation. For example, it has been found that stim-uli that are believed to be potentially harmful due to prior information, are likely to be rated as more painful (Wiech et al., 2010). Since expectations influence how a stimulus is perceived, Wiech and colleagues (2014) tested whether prior information regarding pain leads to a bias in percep-tual decision-making. In their study, participants received explicit instructions regarding cue-shock contingencies. Par-ticipants were presented with neutral geometrical cues (e.g., a circle), indicating a certain probability of receiving either
Figure 1. The visual depiction of the experimental procedure
a high intensity or low intensity electric shock. For exam-ple, a square would indicate 80% probability of receiving a high intensity stimulus and 20% probability of a low in-tensity stimulus. Upon the presentation of the cue and the subsequent administration of shock, participants had to indi-cate, as quickly as possible, whether they had received a high intensity or low intensity shock. A Bayesian hierarchical dif-fusion model (Wiech et al., 2014) was used to analyze the effect of cues on sensory processing and perceptual decision-making. The diffusion model is a cognitive processing model that is used for two-choice decisions (see Data Analysis sec-tion for more informasec-tion). Wiech et al., (2014) found a shift in starting point (z), an indicator of an a priori bias, based upon the cue condition that indicates biased percep-tual decision-making towards the expected sensation. Fur-ther, changes in drift rate (v), which is the mean rate at which sensory information accumulates over time towards the cor-rect or incorcor-rect response, were found to differ depending upon the shock intensity implying altered sensory perception (Wiech et al., 2014). Taken together, these results suggest that in the presence of prior knowledge, incoming sensory information is not analyzed anew but interpreted based on the prior information, leading to altered sensory processing and, thereby,perceptual decision-making.
Experience and Perceptual Processes in Pain
Apart from prior information, an individual’s decision-making processes are also influenced by experience. Re-search on associative learning has reported that informa-tion accumulainforma-tion, through experience, leads to changes in perceptual discriminability of sensory information (Li et al, 2008), enhancing an individual’s ability to discriminate be-tween two previously-indistinguishable stimuli (Ahs et al, 2013). Aizenberg and colleagues (2013), for example, found
that higher learning specificity led to more perceptual dis-crimination acuity, such that the more specific the associa-tive learning process is to a stimulus, the better an individ-ual is at discriminating those stimuli. Further, it has been suggested that perceptual discriminability is influenced by aversive conditioning in pain because perception may be bi-ased by fear (Zaman et al., 2015). Though most of the re-search conducted so far has indicated that associative learn-ing processes or experience leads to enhanced sensory dis-criminability, they have mostly been restricted to olfactory systems and not much work has been done in the field of pain.
The Current Study
To date, research has demonstrated the effect of prior information on perceptual decision-making in pain as ex-plained above. Moreover, the effect of experience on percep-tual decision-making, especially in case of odors, has been well-documented. However, people with chronic pain tend to be reluctant to alter their pain beliefs even when experience tells them otherwise (Tracey, 2010). Therefore, it is impor-tant to understand if, and how, prior information and experi-ence interactively impact perceptual decision-making. Since in previous studies prior information about pain was congru-ent with the experience of pain, it is not clear whether the biases in perceptual decision-making originated due to prior knowledge or experience. The current study aims to inves-tigate the effect of prior instructions about informative cues and differing experience on perceptual decision-making.
It is hypothesized that prior information congruent with experience will lead to altered perceptual decision-making and sensory processing in pain. Meanwhile, when prior information is incongruent with experience, perceptual decision-making could be biased either for or against the
PERCEPTUAL DECISION MAKING IN PAIN 3
given information. It is expected that when participants ex-perience shock intensity congruent to the stimulus cues they receive, they will display a shift in the starting point (z) and changes in drift rate (v). Specifically, in line with Wiech et al. (2014), it is expected that starting point will shift to-wards answering in favor of high pain and drift rate will al-ter based upon shock intensities. Further, it is expected that when participants experience shock intensity incongruent to the stimulus cues they receive, they will demonstrate con-trasting changes in drift rate (v), and the starting point (z) will shift for the low intensity stimulus contingency and will not shift for the high intensity stimulus contingency.
Method Participants
A total of 33 adults (18 Female, average age= 23.9 years) took part in the study.Participants were recruited through the university laboratory website (lab.uva.nl). Participants were compensated with either course credit, 1.5 ec, or e15 for their participation. Participants underwent a phone interview before they were recruited for the experiment. The interview questions pertained to the following: age (≥ 18 years), physi-cal health (heart problems, medication, pregnancy) and men-tal health (anxiety disorders, psychiatric disorders, epilepsy, ADHD, post-traumatic stress disorder), hearing ability, color blindness, and attentiveness, memory and responsiveness. Participants were accepted for the experiment if they met the following selection criteria: no history of heart problems, no history of psychiatric disorders, not pregnant, and have not consumed any drugs within the past 24 hours.
Materials
Stimuli & Apparatus. The experiment employed the use of two commercial electric stimulation devices (Con-stant Current Stimulator, model DS7A; Digitimer©, Hert-fordshire, UK) delivering a train of 1ms monopolar square waveform pulses via a concentric silver chloride electrode. In line with Wiech et al. (2014), stimulation intensities rated as 2 on a Visual Analogue Scale ranging from 0 (= no pain) to 10 (= unbearable pain) were used as an indication of low pain and stimulation intensities rated as 8 were used as an indicator of high pain.
Questionnaires. The participants completed the State-Trait Anxiety Inventory (STAI), Dutch version, which con-sists of 20 questions for assessing the state anxiety and 20 questions for trait anxiety. Internal consistency coefficients for the scale have ranged from .85 to .95 (van der Ploeg, 1985); test-retest reliability coefficients for STAI, English version, have ranged from .65 to .75 over a 2-month inter-val (Spielberger, Gorsuch, Lushene, Vagg & Jacobs, 1983). Further, the participants also completed the Anxiety Sensitiv-ity Index-3 (ASI-3) (Dutch version) which is a questionnaire
consisting of 16 items, with internal consistency coefficients ranging from .43 to .83 (Vujanovic et al., 2007). Lastly, par-ticipants completed the Pain Sensitivity Questionnaire (PSQ) which consists of 17 items and has internal consistency co-efficients ranging between 0.52 and 0.71 (Ruscheweyh et al., 2012). The current study will not be analyzing the data ob-tained from questionnaires.
Procedure
The participants provided informed consent and com-pleted the STAI-S questionnaire. (see Figure 1)
Preparation.
1. Calibration Procedure & Discrimination Test
The participants underwent a calibration procedure to determine the level of low intensity and high intensity stimulations. The stimulus equipment was attached on the wrist of participants’ non-preferred hand and they received incrementing shocks that they had to rate on a scale from 0-10. In line with Wiech et al. (2014), a rating of 2 was classified as a low intensity stimu-lus and 8 was a high intensity stimustimu-lus. A total of two rounds of calibration were performed to determine the accurate levels of shock intensity. After the calibration procedure, participants performed a discrimination test wherein they received two high intensity and two low intensity stimulations according to their calibration lev-els. If a participant successfully identified both the tri-als, he/she proceeded to the next task. If not, the partici-pant repeated the calibration procedure and the discrim-ination test. All participants were able to successfully complete the discrimination test within 2 trials. (see Figure 1)
2. Cue Recognition Test
Participants completed a cue recognition task that tested their understanding of visual cue-shock stimulation contingencies. They were presented with one of two visual cues, a square or a circle (a white outline of the shapes). One cue signaled the 80% probability of re-ceiving a high intensity shock and 20% probability of a low intensity shock (80/20 condition). The other cue signaled equal probability of receiving a high or low intensity stimulus (50/50 condition). Participants were presented with the cues ten times and they had to in-dicate if the cue signaled the probability of receiving a high intensity stimulus 80% or 50% of the time. Be-fore performing the cue recognition task, participants were provided with explicit instructions regarding the cue contingencies. Only if a participant correctly rec-ognized the cue contingency at least 80% of the time, he/she was allowed to proceed to the main task. Par-ticipants had two chances to successfully complete the
Figure 2. The visualization of a trial for the Reaction Time Task. The task begins with the appearance of a fixation cross on the screen moving on to the presentation of the cue (picture of square or circle) according the cue condition assigned to the participant. Then the participants receive the shock stimulus (high or low) according to whether the block is congruent or incongruent. The participants then make their response.
cue recognition task, failing which they were dismissed from the study. (see Figure 1)
Reaction Time Task. The main task of the experiment involved the presentation of a visual cue and three seconds later, the application of a high or low intensity shock. Partic-ipants had to indicate, as quickly as possible, whether they had received a high intensity or low intensity stimulus. Par-ticipants first received cue instructions, after which they ex-perienced 4 blocks of stimulus intensities congruent with the provided information and 4 blocks of stimulus intensities in-congruent with the provided information. Each participant underwent 8 blocks of 28 trials each and after each block they provided ratings for average intensity of pain and expe-rienced anxiety. Before the main blocks began, participants performed 16 practice trials, 8 for high intensity and 8 for low intensity. In total, all participants received 224 trials for the main task (8 blocks with 28 trials each and 16 practice trials). (see Figure 2)
Exit Interview & Questionnaires. All participants completed the State-Trait Anxiety Inventory (STAI), Anxiety Sensitivity Index-3 (ASI) and Pain Sensitivity Questionnaire (PSQ), and provided demographic information (such as sex, age, nationality etc.) at the end of the experiment. Further, the participants answered questions about the pain they ex-perienced (pleasantness and intensity of stimulus, and how much they were startled by the stimulus), and the believabil-ity of the provided instructions.
Data Analysis
First, the differences in RTs (Response Times) and accu-racies between the different conditions were computed using pairwise t-tests (Bonferroni correction was used). Further, a factorial ANOVA was conducted for a within-subjects de-sign with the independent factors: Cue condition (80/20 or 50/50) and Shock intensity (high or low). Data analysis uti-lizes the Bayesian hierarchical diffusion model. Diffusion model assumes that “decisions are made by a noisy process that accumulates information over time from a starting point toward one of two response criteria or boundaries” (Ratcliff & McKoon, 2008, p.875). The parameters that are estimated are starting point (z), drift rate (v), boundary separation (a), and non-decision time (t). A bias in starting point (z), an a prioribias, would indicate the expectation of a certain per-cept, thereby shifting towards the corresponding boundary, and leading to short response times and increased accuracy on correct trials. On the other hand, a bias in drift rate (v) is suggested by accelerated evidence accumulation towards the boundary. Boundary separation (a) refers to the evidence re-quired to make a decision and non decision time (t) includes the remaining parts of the decision process (e.g., memory retrieval). In the model that the current study employs, the drift rate is allowed to differ across all conditions while the starting point differs according to task instruction or cue con-tingency. Non-decision time and boundary separation were fixed across all conditions. Prior distributions of the param-eters, reflecting prior knowledge, were established. Statisti-cal analysis is based upon compared contrasts of tail ends of posterior distributions, which represent knowledge of
param-PERCEPTUAL DECISION MAKING IN PAIN 5
Figure 3. Mean response times for correct and incorrect responses, and accuracy rates for both congruent and incongruent blocks
eters taking into account the data. Comparisons were made between starting point (z) of the 80/20 and 50/50 cue con-dition of the congruent contingency. For example, to test whether the a priori bias parameter was closer to the up-per boundary for 80/20 or 50/50 condition, we computed the probability (z80/20-z50/50) ≤ 0. If the probability was less than
2.5% then we assumed that z80/20 > z50/50, which means that
the response is biased towards high the 80/20 cue condition, and if probability was greater than 97.5% then the response was considered to be biased towards the 50/50 cue. If the probability was between 2.5% and 97.5% then judgment was suspended. Similarly, comparisons were made between the starting point (z) of the 80/20 cue condition for congruent and incongruent contingencies, and the starting point (z) of the 50/50 cue condition for congruent and incongruent con-tingencies. Similar comparisons were conducted for the drift rate (v) parameter.
Results Decision Accuracies& Response Times
Accuracy rates were higher for high intensity pain for both 80/20 condition (t(31) = -3.33, p<0.001) and 50/50 cue con-dition (t(31) = -2.81, p= 0.005). Participants were faster in indicating their response for high intensity regardless of the cue condition or the congruency with prior instructions (t(31)= -1.165, p= 0.24) (see Figure 3). Response times for low intensity did not differ across cue condition (t(31) = 1.05, p= 0.3), and were slower in both cue conditions and across
the blocks. These results suggest that response times, slow or fast, and the accuracy of recognition, high or low intensity, depended upon the intensity of shock being received and not the cue, square or circle, being presented. The results in the incongruent blocks showed the most errors and the slowest response times in recognition of low intensity (t(31)= 1.06, p= 0.3). This might indicate that participants could not rec-oncile for the incongruency between prior instructions and opposing experience which caused more errors. This could also suggest that participants experienced habituation which made it difficult for them to correctly recognize low inten-sity shocks. A factorial analysis of the variance of response times found a significant effect of shock intensity (F = 12.30, p< 0.001, d = 0.85), meanwhile cue condition did not yield a significant effect (F = 0.39, p = 0.54, d = 0.15). These results indicate that variability in response times depended upon shock intensity and the cue condition did not have any effect. No significant results were found for an interaction of cue condition and shock intensity, as well, which suggests that only shock intensity determined the level of accuracy and speed of response times. Further, the data from the ques-tionnaires yielded the following results:
Mean Standard Deviation Standard Error
ASI 13.85 5.61 0.98
STAI-T 39.30 11.21 1.95
STAI-S 33.76 8.14 1.42
Figure 4. Drift rate& starting point parameters for congruent & incongruent blocks
The results from the questionnaire data fell within 3 stan-dard deviations of the mean and displayed no extreme values. These suggest that none of the participants displayed abnor-mal levels of trait or state anxiety, nor were they unusually sensitive to anxiety or pain.
Diffusion Model
The current model refers only to the a-priori bias and drift rate parameters since these are the two parameters that con-cern the subject of the study. The MCMCs (Monte Carlo Markov Chain) were checked visually and they reached con-vergence.
Starting Point. The analysis of planned contrasts of starting point posteriors did not yield significant results: p(z50/50:80/20) = 99.95 (congruent condition) & p(z50/50:80/20)
= 21.45 (incongruent condition). An analysis of both the congruent and incongruent conditions shows that mean start-ing point for low and high intensity are very similar to each other. For the congruent condition (experience in line with prior instructions), the mean starting point for cue condition 80/20 was 0.52 and mean starting point for cue condition 50/50 was 0.51 (see Figure 4). For the incongruent condi-tion (experience in contrast with prior instruccondi-tions), the mean starting point for both cue conditions 80/20 and 50/50 was equal to 0.53. These results suggest that the responses of participants were not biased according to the cue condition, such that their decisions were not biased by prior instructions about the cues and shock intensities.
Drift Rate. The analysis of mean drift rate yields higher drift rate for high intensity (cue 80/20 = 2.89 and cue 50/50 = 2.74) and lower drift rate for low intensity (cue 80/20 = 2.42
and cue 50/50 = 2.84) across both cue conditions in congru-ent blocks. A similar pattern was observed in incongrucongru-ent blocks wherein the high intensity yielded high drift rates (cue 80/20 = 2.29 and cue 50/50 = 2.39) and low intensity yielded low drift rates (cue 80/20 = 1.90 and cue 50/50 = 1.84) across both cue conditions. Further, the planned contrasts of drift rate reached significance for shock intensity, p(vhigh:low)= 0.
This indicates a main effect of stimulation intensity such that the process of information accumulation was biased towards the high intensity shock. The planned contrasts of drift rate for cue condition failed to reach significance, p(v80/20:50/50)=
14.72. The planned contrasts of drift rate for the interaction (cue condition X stimulation intensity) also yielded signifi-cant results, p(v50/50:80/20)= 0, indicating that the process of
decision-making is influenced by the speed of information accumulation and a priori bias. The interaction effect can be seen particularly for the drift rate for the low intensity shock in the cue condition 50/50 during the congruent blocks (Fig-ure 4), where the drift rate increases for the low intensity stimuli in the condition. Overall, the results from the drift rate analysis suggest that the process of information accu-mulation to reach a decision is affected by a combination of both the shock intensity experienced and prior information about the cue presented.
Discussion
We hypothesized that perceptual decision-making regard-ing the intensity of a pain stimulus will be altered in the pres-ence of prior information about the cue-shock contingencies. The results from the diffusion model analysis show no dif-ference in starting point (z) between the 80/20 and the 50/50 cue condition for the congruent contingency (i.e. instruc-tions were congruent with pain experienced). This finding is in contrast to the results of Wiech et al. (2014), who found
PERCEPTUAL DECISION MAKING IN PAIN 7
that after participants were instructed about the cue-shock contingencies, the starting point (z) shifted towards the ex-pected sensation. Further, starting point (z) values remained the same also for the incongruent contingency (i.e. prior in-structions were in contrast to the pain experienced). Taken together, results suggest that prior instructions do not induce a bias in the perceptual decision-making process in pain. On the other hand, the drift rate analyses show that sensory pro-cessing was altered according to the stimulus intensity (high or low shock) and cue condition (80/20 and 50/50), in line with the finding of Wiech et al. (2014). Regardless of the cue condition or congruency with prior instructions, the drift rate was higher for high intensity and lower for low intensity shocks. But the drift rate posterior analyses show the pres-ence of an interaction effect of stimulus intensity and cue condition, in contrast with Wiech et al., (2014). The inter-action effect is evident in the high drift rate for low intensity in the 50/50 cue condition in the blocks congruent with prior instructions. This points out that speed of information accu-mulation was higher for low intensity shocks for the 50/50 cue condition compared to the 80/20 cue condition. Further, this effect can be seen in the incongruent blocks where high intensity shocks are more quickly recognized in the 50/50 cue condition, as opposed to the 80/20 cue condition.
Our findings about prior instructions show that previous information about pain and the experience of pain do not seem to influence the a-priori bias in painful situations. This means that even though individuals have information about the probability of pain stimulus intensity they might receive, they are not using the information to make their decisions. This is in contrast with previous literature (Wiech et al., 2014) according to which expectations regarding pain tend to modulate the decisions about pain or pain relief an individual experiences.
This contrast in results between the current and the previ-ous study (Wiech et al., 2014) could be attributed to the use of the hierarchical diffusion model. The hierarchical diffu-sion model has been successfully applied to explain percep-tual decision-making in the past but it has been restricted to visual decision-making tasks, such as discrimination based upon brightness or numerosity (Ratcliff & Rouder, 1998), the Stroop task (Stroop, 1935; Williams et al., 1996) and the dot-probe task (MacLeod et al., 1986; Salemink, Hout & Kindt, 2007). Wiech et al. (2014) was the first study, to our knowl-edge, that applied hierarchical diffusion model to a percep-tual decision-making task that involved the physical applica-tion of the pain stimuli. Since Wiech et al. (2014) is the first study to do so and the current study is the second, there is a need for more data to examine the boundary conditions and parameters. This points to a need for further replication and examination of perceptual decision-making in pain wherein the pain stimulus is physically applied. Another reason for the contrast could be experimenter effects, specifically racial
and gender effects, during data collection. Studies investigat-ing experimenter effects have found that participants provide more desirable responses and perform better when the exper-imenter is of a different race and gender (Bradley, Snyder & Katahan, 1972; Sattler, 1970; Rosenthal, 1966). Due to these experimenter effects, it could be that the participants chose to ignore their bias for the cues and provided answers which would be most desirable, thereby, reducing the effect of a pri-oribias created by the prior instructions. Further, the contrast in results could also indicate the presence of hidden modera-tors in the relationship between prior information and biased perceptual decision-making. This notion is further supported by the presence of an interaction effect of prior information and experience for the drift rate parameter. Lastly, the con-trast could also be a result of participant fatigue. The partici-pants had to undergo a lot of trials throughout the experiment which could have caused habituation to the shock sensations, despite calibration after every block. The habituation could have lead to difficulties with processing the intensity of the shock leading to slower responses as opposed to the more quick and automatic responses that an a priori bias is likely to produce.
For incongruent blocks, it could be that the participants quickly learned that the cue condition did not follow up with a shock of appropriate intensity. Therefore, they chose to use only their experience because they learned that the prior information was differing from their experience. This finding indicates that individuals are likely to value their experience more during decision-making about pain, despite the prior in-formation provided. This suggests that fear learning through experienced CS-US contingencies is more likely to impact perceptual decision-making in pain situations than prior in-formation.
The results, also, indicate that sensory processing for de-cisions about pain intensity are impacted by a combination of experience and prior knowledge about pain. This finding suggests that the current experience of pain is weighed in with the expectations about pain in order to decide whether the pain being felt is high or low. The process of decision-making about pain utilizes all the knowledge, prior informa-tion and own experience, available to recognize the pain be-ing felt. In context of chronic pain, decisions about pain may be made based upon the pain the individual is experiencing and the prior knowledge, e.g. previous incidents of pain, that an individual possesses. This finding could be studied further and it might be investigated if there exists a possible modera-tion of the relamodera-tionship between experience and pain-related decisions by prior knowledge.
Future studies need to, firstly, conduct direct replications of the findings of the current study to establish a valid and re-liable effect. Further, studies can use more fear-relevant stim-uli, such as pictures of threatening objects or animals, instead of using geometrical shapes to induce a priori bias. This is
likely to produce a stronger effect of a priori bias because studies have indicated that threatening stimuli (e.g. snakes, guns) tend to attract more attention than neutral stimuli (e.g. shapes, househould objects) (Fox, Griggs & Mouchlianitis, 2007; Brosch & Sharma, 2005). Since, threatening pictures capture attention easily, they are likely to be paired easily with the threatening shocks, creating a stronger bias which could be reflected in perceptual decision-making. Further, the current study employs only the starting point and drift rate parameters of the hierarchical diffusion model. Future research can apply the full model and can utilize other pa-rameter, such as non-decision time, to better explain percep-tual decision-making.
In conclusion, our results indicate that prior knowledge about pain intensity does not seem to create any bias towards the expected intensity in perceptual decision-making. Fur-ther, the process of reaching a decision seems to be impacted by a combination of both prior knowledge and experience.
References
[1] Aizenberg, M. & Geffen, M.N., 2013. Bidirectional ef-fects of aversive learning on perceptual acuity are me-diated by the sensory cortex. Nature Neuroscience, 16, 994-996.
[2] Apkarian, A. V., Hasmi, A. J. & Baliki, M. N. (2011). Pain and the brain: Specificity and plasticity of the brain in clinical chronic pain.Pain, 152, S49-S64. [3] Ahs, F., Miller, S. S., Gordon, A. R. & Lundstrom, J.
N. (2013). Aversive learning increases sensory detec-tion sensitivity. Biological Psychology, 92, 135-141. [4] Bradley, N., Snyder, C. R. & Katahan, M. (1972). The
effects of subject race and sex and experimenter race upon classroom-related risk-taking behavior. Psycho-nomic Science, 28, 362-364.
[5] Brosch, T. & Sharma, D. The role of fear-relevant stim-uli in visual search: a comparison of phylogenetic and ontogenetic stimuli. Emotion, 5, 360-364.
[6] De Ridder, D., Elgoyhen, A. B., Romo, R. & Langguth, B. (2011).Phantom percepts: Tinnitus and pain as per-sisting aversive memory networks. PNAS, 108, 8075-8080.
[7] Dunovan, K. E., Tremel, J. J. & Wheeler, M. E. (2014). Prior probability and feature predictability interactively bias perceptual decisions. Neuropsychologia, 61, 210-221.
[8] Fox, E., Griggs, L. & Mouchlianitis, E. (2007). The De-tection of Fear-Relevant Stimuli: Are Guns Noticed as Quickly as Snakes? Emotion, 7, 691-696.
[9] Hart, R.P., Wade, J.B. & Martelli, M.F. (2003). Cog-nitive impairment in patients with chronic pain: the significance of stress. Current Pain& Headache Re-ports,7, 116-126.
[10] Heekeren, H. R., Marrett, S., Bandettini, P. A. & Unger-lelder, L. G. (2004). A General Mechanism for Percep-tual Decision Making in the Human Brain. Nature, 431, 859-862.
[11] Karim, M., Harris, J. A., Morley, J. W. & Breakspear, M. (2011). Prior and Present Evidence: How Prior Ex-perience Interacts with Present Information in a Percep-tual Decision Making Task. PLoS ONE, 7, e37580. [12] Li, W., Howard, J. D., Parrish, T. B. & Gottfried, J.
A. (2008). Aversive Learning Enhances Perceptual and Cortical Discrimination of Indiscriminable Odor Cues. Science, 319, 1842-1845.
[13] MacLeod, C., Mathews, A., & Tata, P. (1986). Atten-tional bias in emoAtten-tional disorders. Journal of Abnormal Psychology, 95, 15-20.
[14] Ploeg, H. M. van der, Defares, P. B. & Spielberger, C. D. Zelf Beordelings Vragenlijst STAI versie DYI en versie DYII. Lisse: Swets en Zietlinger. 1979
[15] Ploeg, H. M. van der. (1985). The development and validation of of the Dutch State-Trait Anxiety Inven-tory: Zelf Beordelings Vragenlijst. Stress and Anxiety, 9, 129-138.
[16] Ratcliff, R. & McKoon, G. (2008). The Diffusion Deci-sion Model: Theory and Data for Two-Choice DeciDeci-sion Tasks. Neural Computing& Applications, 20, 873-922. [17] Ratcliff, R., & Rouder, J. N. (1998). Modeling response times for two-choice decisions. Psychological Science, 9, 347-356.
[18] Rosenthal, R. (1966). Experimenter effects in behavior research. Behavioral Research.
[19] Ruscheweyh, R., Verneuer, B., Dany, K., Marziniak, M., Wolowski, A., Colak-Ekici, R. & Knecht, S. (2012). Validation of the pain sensitivity questionnaire in chronic pain patients. Pain, 153, 1210-1218. [20] Salemink, E., Hout, M. A. van den & Kindt, M. (2007).
Selective attention and threat: Quick orienting versus slow disengagement and two versions of the dot probe task. Behaviour Research& Therapy, 45, 607-615. [21] Sattler, J.M. (1970). Racial "experimenter effects" in
experimentation, testing, interviewing, and psychother-apy. Psychological Bulletin, 73, 137-160.
PERCEPTUAL DECISION MAKING IN PAIN 9
[22] Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the State Trait Anxiety Inventory.
[23] Stewart-Williams, S & Podd, J. (2004). The placebo effect: Dissolving the expectancy versus conditioning debate. Psychological Bulletin, 130, 324-340.
[24] Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662.
[25] Taylor, S., Zvolensky, M. J., Cox, B. J., Deacon, B., Heimberg, R. G., Ledley, D. R.,Abramowitz, J. S.,..., Cardenas, S. J. (2007). Robust Dimensions of Anxiety Sensitivity: Development and Initial Validation of the Anxiety Sensitivity Index 3. Psychological Assessment, 19, 176-188.
[26] Tracey, I. (2010). Getting the pain you expect: mecha-nisms of placebo, nocebo and reappraisal effects in hu-mans. Nature Medicine, 16, 1277-1283.
[27] Vujanovic, A. A., Arrindell, W. A., Bernstein, A., Nor-ton, P. J. & Zvolensky, M. J. (2007). Sixteen-Item Anxiety-Sensitivity Index: Confirmatory Factor Ana-lytic Evidence, Internal Consistency and Construct
Va-lidity in a Young Adult Sample from the Netherlands. Assessment, 14, 129-143.
[28] Wiech, K., Lin, C. S., Brodersen, K. H., Bingel, U., Ploner, M., & Tracey, I. (2010). Anterior insula inte-grates information about salience into perceptual de-cisions about pain. The Journal of Neuroscience, 30, 16324-16331.
[29] Wiech, K., Vandekerckhove, J., Zaman, J., Tuerlinckx, F., Vlaeyen, J. W. S., Tracey, I. (2014).Influence of prior information on pain involves biased perceptual decision-making. Current Biology, 24, R679-R681.
[30] Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathol-ogy. Psychological Bulletin, 120, 3-24.
[31] Zaman, J., Vlaeyen, J. W. S., van Oudenhove, L., Wiech, K. & van Diest, I. (2015). Associative fear learning and perceptual discrimination: A perceptual-pathway in the development of chronic pain. Neuro-science and Biobehavioral Reviews, 51, 118-125.
