I am a
Home I AM A Search Login

Changing pain thresholds with classical conditioning



The 2024 Global Year will examine what is known about sex and gender differences in pain perception and modulation and address sex-and gender-related disparities in both the research and treatment of pain.

Learn More >

Our previous post about a classical conditioning model for pain generated some lively discussion.  Some argued vehemently that pain cannot be a classically conditioned response, and others argued vehemently that of course it can and we have known this for decades.  We haven’t yet pinned our colours to any particular mast – we’d like to see the science speak for itself.  However, we’ve just taken a big step forward.

The classical conditioning model for pain suggests that when non-noxious input (e.g. a movement) is paired with noxious and pain-eliciting input (e.g. nociceptive signalling) during the acute phase of an injury, it is possible for the non-noxious input (the movement) to become capable of eliciting pain – in the absence of the noxious input.  In other words, after tissue healing, the movement could elicit pain without any nociceptive activity.

This idea is popular amongst clinicians, but contrary to popular belief (see Madden & Moseley 2016), it’s not yet cut and dried, and there are several questions we still have to ask in the laboratory.  One of those questions is whether classical conditioning can change pain threshold.  Williams and Rhudy (2007) have previously demonstrated that changes in arousal or valence are able to modulate the pain threshold (the term ‘valence’ refers to ‘like’ or ‘dislike’ for something) – and classical conditioning procedures are often used to induce changes in arousal or valence.  Williams and Rhudy showed that a participant who is exposed to repeated pairing of non-noxious stimulus A before a noxious (and painful) stimulus grows to dislike non-noxious stimulus A.  If that participant is then exposed to A just before a pain threshold test, the pain threshold is lower than when tested without A.

We call that an indirect modulation of pain threshold, and it might just reflect expectation because the disliked non-noxious stimulus was always presented before the noxious stimulus or the test stimulus.  We know, from the placebo/nocebo literature, that expectation is a powerful modulator of pain.  In the clinical context, a person who has recognised the triggers for their pain (e.g. certain movements) expects pain with those triggers, so they tend to avoid those triggers.  Very sensible, we think.  However, in many cases, those triggers (e.g. certain movements) are not inherently dangerous – so perhaps this avoidance strategy is not as sensible as it first appears.  The situation we are most interested in is when someone has not recognised the triggers for their pain, and so continues to expose themselves to the triggers.  This repeated exposure with pain constitutes a repeated pairing of the trigger with noxious input to the brain, repeatedly eliciting pain during the acute phase of tissue injury.  Because the person doesn’t recognise the trigger as linked to their pain, there is no element of anticipation or expectation here.  In this situation, does the repeated pairing of the trigger with the nociceptive input (plus a response of pain each time) lead to a situation in which the trigger itself becomes capable of modulating the pain threshold?  In other words, if anticipation and expectation are removed, can a non-noxious stimulus still modulate the pain threshold?

In order to ask this question, we had to cut out the time delay between the non-noxious (neutral) stimulus and the noxious stimulus.  This is an unconventional (though not unique) move: most classical conditioning studies have the neutral (informational) stimulus precede the response-evoking stimulus.

We delineated three locations on each participant’s back.  The middle location was a rectangular area for laser stimulation. Above and below the laser rectangle, we attached modified mobile phone vibrators to the skin, so that we had two locations for identical vibratory stimulation and one ‘zone’ for laser stimulation.  Then we began a pairing procedure.  Each time we activated vibrator 1, we also delivered a very painful laser stimulus.  Each time we activated vibrator 2, we delivered a non-painful laser stimulus.  We did that many times.  Then we tested the results.  In the test phase, each time we activated either vibrator 1 OR vibrator 2, we delivered a laser stimulus that was calibrated to match the participant’s pain threshold, as tested before the pairing procedure.  A truly at-pain-threshold stimulus should be experienced as painful in 50% of trials and as non-painful in 50% of trials.  We found that participants rated at-pain-threshold trials that were paired with vibrator 1 (previously paired with very painful laser stimuli) as painful more than 50% of the time, and they also rated at-pain-threshold trials that were paired with vibrator 2 (previously paired with non-painful laser stimuli) as painful less than 50% of the time.

This indicates that ‘pairing history’ can modulate the likelihood of experiencing pain to nociceptive stimuli that are around pain threshold.  If we frame this using the popular (and, I would argue, evidence-based) idea that pain depends on an appraisal of threat, we might interpret this by saying that the pairing experience endows the non-noxious (in this case, vibratory) stimulus with the ability to shift the balance of appraisal towards or away from the conclusion that there is a body part under threat that should be protected.  In this study, we found no difference in the way participants rated the vibratory stimuli when they were delivered alone, implying that their influence on how the combined vibration-laser trials were experienced was a modulatory effect, rather than something that could be attributed to a change in the way the vibrations themselves were perceived.

So it looks as though you don’t need the opportunity to anticipate pain in order to have a lowered pain threshold: your system can modify its appraisal of threat and, consequently, your perceptual experience, in real time.  The patient who hasn’t recognised his triggers may experience pain more easily when he exposes himself to them.  That could be useful in the sub-acute phase of an injury, because it might motivate him to learn to recognise the triggers and avoid them next time, which would allow for tissue healing.  But if that lowered pain threshold persists after tissue has healed, then he might be heading into painful territory… Of course, we don’t yet have evidence for that, but we are working on it. Watch this space!

PS: As with all studies showing a new finding, this needs to be replicated.  And we’re working on it…

Tory Madden

Tory MaddenTory is a physiotherapist who worked clinically before turning her focus toward research.  She did her PhD with BiM, looking at classical conditioning and pain.  She is now a postdoc at the University of Cape Town, where she is exploring the role of social threat and is also involved in teaching.  She is an associate editor for BiM.

Tory spends some of her spare time running in the mountains – a necessary pastime that conveniently balances her penchant for excessive consumption of chocolate and other tasty delights.


Victoria J. Madden, Valeria Bellan, Leslie N. Russek, Danny Camfferman, Johan W. S. Vlaeyen, and G. Lorimer Moseley (2016). Pain by Association? Experimental Modulation of Human Pain Thresholds Using Classical Conditioning. J Pain 17 (10), 1105-1115

Madden, V. J. and G. L. Moseley “Do clinicians think that pain can be a classically conditioned response to a non-noxious stimulus?” Manual Therapydoi:10.1016/j.math.2015.12.003

Williams, A.E. & Rhudy, J.L. (2007) The influence of conditioned fear on human pain thresholds: does preparedness play a role? J Pain, 8, 598-606.

Share this