Background Information


  1. An article from Simmons Foundation/Autism Research Initiative
  2. An online report by Randy Mertens from University of Missouri
  3. A video report from KSL-TV

Summary of research findings

Pupillary light reflex (PLR)
PLR refers to the involuntary response whereby the pupil size changes in response to a short flash light.

Using a binocular pupillogram system (Fan et al., 2009a; 2009b), we examined PLR in 44 children of typical development (age 10.4 ± 2.7 years, range 6-16 years, 23 females and 21 males), and 24 children with ASDs (age 12.9 ± 4.3 years, range 7-20 years, 22 males and 2 females). The control group had normal cognitive development and was progressing in age-appropriate educational programs. Participants with ASD met DSM – IV diagnostic criteria on the Autism Diagnostic Interview-Revised (ADI-R) and/or the Autism Diagnostic Observation Schedule, which was substantiated by an in depth clinical interview and evaluation. The ASD group included 10 children with autistic disorder, 10 with Asperger’s syndrome, and 4 with Pervasive Developmental Disorder-Not Otherwise Specified (PDD-NOS).


Fig.1. An illustration of the pupillogram and extracted PLR parameters. Please see text for explanations of each parameter.

The participant’s eye was stimulated by a brief (100 ms) light flash (530nm). We measured PLR in both light-adapted (LA) and dark-adapted (DA) conditions. The LA test was conducted in well-lit room with a stimulus intensity of LA 872 cd/m2. The room lights were turned off after the LA tests. The participants stayed in the dark room for 20 minutes to adapt to the darkness and then the tests were repeated by applying two stimulus intensities of 0.5 cd/m2 and 794 cd/m2. During each single test, the participant was instructed to look at cartoon pictures on a computer screen 4 feet away and their pupil images were acquired by using infrared cameras.

The acquired pupil images were automatically processed to extract the pupillograms (change in pupil size with time shown in Fig.1). As illustrated in Fig. 1, the following basic PLR parameters were measured for each measurement: the initial pupil diameter D0, the maximal constriction diameter Dm, the PLR latency tL (interval between stimulus onset at time zero and beginning of constriction), the constriction time tC (interval between constriction onset and the maximal constriction), and the recovery time tR (time interval between the maximal constriction and the recovery to half of the maximal constriction amplitude). The relative constriction (in percentage) was calculated as (D02-Dm2)/D02. The constriction velocity and the recovery velocity were calculated as (D0-Dm)/tC and (D0-Dm)/2tR, respectively.

Preliminary findings
Children with an ASD exhibited significantly longer PLR latencies (see Fig. 1 for definition). Figure 2a shows the distribution of latencies measured at both dark-adapted (DA) 794 cd/m2 and light-adapted (LA) 872 cd/m2. ANOVAs (Fig. 2b) confirm that latencies are significantly different between the two groups (p < 0.0001) under both DA 794 cd/m2 and LA 872 cd/m2. The p-values were Bonferroni corrected. At a lower light stimulation intensity (DA 0.5 cd/m2), the difference remains significant at p=0.0016. This slightly higher p-value is due to larger measurement variations associated with smaller pupil constrictions at low stimulus.
Fig. 2. (a) Distribution of PLR latency and (b) statistics of the mean latency in the ASD and control groups. Data shown here were from 44 typically developing children and 24 children with ASD. The error bars in (b) indicate the standard deviation. (c) A linear classification using Perceptron algorithm shows that 100% classification was achieved by incorporating PLR latency and relative constriction.

The autism group also exhibited smaller pupillary constriction amplitudes and lower constriction velocity than typically developing participants. There was no significant difference in mean values of other PLR parameters (p > 0.1). However, F statistics indicated that the variance in pupil diameters of autism was significantly larger at both light and dark adaptation (p < 0.0001 in both instances). Similarly, large variations were also observed in PLR latency in the autism group (Fig.1a). These may be an indication of heterogeneity.

Figure 2(c) shows that the Perceptron algorithm can completely separate the autism group from the controls by using two PLR parameters: relative constriction and latency. In addition, using the discriminant procedure DISCRIM (implemented within the SAS statistical software package) with cross-validation (leave-one-out algorithm), 92.5% of participants were successfully classified, with 7.0% false positives (three controls misclassified) and 8.3% false negatives (two individuals with ASD misclassified).


X.F. Fan, J.H. Miles, N. Takahashi, and G. Yao, “Abnormal transient pupillary light reflex in individuals with autism,” J. Autism and Developmental Disorders 39, 1499-1508 (2009).

X.F. Fan, L. Hearne, B. Lei, J.H. Miles, N. Takahashi, G. Yao, “Weak gender effects on transient pupillary light reflex,” Autonomic Neuroscience: Basic and Clinical, 147,  9-13 (2009).
X.F. Fan, J.H. Miles, N. Takahashi, and G. Yao, “Gender-specific lateralization of constriction anisocoria in transient pupillary light reflex,” Invest Ophthalmol Vis Sci., 50(3), 1137-1144 (2009).

All Rights Reserved Copyright © Biological Optics Lab.2009-2010