No evidence of conditioning of pupillary constriction despite overtraining

Abstract

Eyeblink conditioning is the most popular paradigm for studying classical conditioning in humans. But the fact that eyelids are under voluntary control means it is ultimately impossible to ascertain whether a blink response is ‘conditioned’ or a timed ‘voluntary’ blink response. In contrast, the pupillary response is an autonomic response, not under voluntary control. By conditioning the pupillary response, one might avoid potential volition-related confounds. Several attempts have been made to condition the pupillary constriction and dilation responses, with the earliest published attempts dating back to the beginning of the 20th century. While a few early studies reported successful conditioning of pupillary constriction, later studies have failed to replicate this. The apparatus for recording pupil size, the type of stimuli used and the interval between the stimuli has varied in previous attempts—which may explain the inconsistent results. Moreover, measuring the pupil size used to be cumbersome compared with today when an eyetracker can continuously measure pupil size non-invasively. Here we used an eyetracker to test whether it is possible to condition the autonomic pupillary constriction response by pairing a tone (CS) and a light (US) with a 1s CS-US interval. Unlike in previous studies, our subjects went through multiple training sessions to ensure that any potential lack of conditioning would not be due to too little training. A total of 10 participants went through 2–12 conditioning sessions, each lasting approximately 20 min. One training session consisted of 75 paired, tone + light, trials and 25 randomly interspersed CS alone trials. The eyetracker (Tobii Pro Nano), continuously measured participants’ pupil size. To test statistically whether conditioning of the pupillary response occurred we compared the pupil size after the tone on the first session and the last session. The results showed a complete lack of evidence of conditioning. Though the pupil size varied slightly between participants, the size did not change as a result of the training—irrespective of the number of training sessions. The data replicate previous findings that pupillary constriction does not show conditioning. We conclude that it is not possible to condition pupillary constriction—at least not by pairing a tone and a light. One hypothesis is that when pupillary conditioning has been observed in previous studies, it has been mediated by conditioning of an emotional response.

Keywords

• Classical conditioning
• Pupillary constriction
• Associative learning
• Timing
• Autonomic reflexes