Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Adults with autism overestimate the volatility of the sensory environment

Nature Neuroscience volume 20, pages 1293–1299 (2017)Cite this article

Subjects

This article has been updated

Abstract

Insistence on sameness and intolerance of change are among the diagnostic criteria for autism spectrum disorder (ASD), but little research has addressed how people with ASD represent and respond to environmental change. Here, behavioral and pupillometric measurements indicated that adults with ASD are less surprised than neurotypical adults when their expectations are violated, and decreased surprise is predictive of greater symptom severity. A hierarchical Bayesian model of learning suggested that in ASD, a tendency to overlearn about volatility in the face of environmental change drives a corresponding reduction in learning about probabilistically aberrant events, thus putatively rendering these events less surprising. Participant-specific modeled estimates of surprise about environmental conditions were linked to pupil size in the ASD group, thus suggesting heightened noradrenergic responsivity in line with compromised neural gain. This study offers insights into the behavioral, algorithmic and physiological mechanisms underlying responses to environmental volatility in ASD.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Task structure.
Figure 2: Behavioral results based on the ground truth.
Figure 3: Relationship between behavioral surprise and symptoms.
Figure 4: Computational-model details and results.
Figure 5: Pupillometry results.

Similar content being viewed by others

Change history

  • 07 August 2017

    In the version of this article initially published online, the first sentence of the Online Methods referred to “29 adults with ASD and 26 healthy NT volunteers.” To avoid any implication that those with ASD are not healthy, this has been changed to “29 adults with ASD and 26 NT volunteers.” Similarly, in the first paragraph of the “Learning-rate data” section, “healthy volunteers” has been changed to “NT volunteers.” In the Life Sciences Reporting Summary, “healthy” has been changed to “NT” in the first sentences of items 3 and 12. In the supplementary information originally posted online, the legend to Supplementary Figure 4 read “non-clinical healthy volunteers.” This has been changed to “non-clinical volunteers.” The errors have been corrected in the PDF and HTML versions of this article.

References

  1. Kenny, L. et al. Which terms should be used to describe autism? Perspectives from the UK autism community. Autism 20, 442–462 (2016).

    PubMed  Google Scholar 

  2. Pellicano, E. & Burr, D. When the world becomes 'too real': a Bayesian explanation of autistic perception. Trends Cogn. Sci. 16, 504–510 (2012).

    PubMed  Google Scholar 

  3. Van de Cruys, S. et al. Precise minds in uncertain worlds: predictive coding in autism. Psychol. Rev. 121, 649–675 (2014).

    PubMed  Google Scholar 

  4. Lawson, R.P., Rees, G. & Friston, K.J. An aberrant precision account of autism. Front. Hum. Neurosci. 8, 302 (2014).

    PubMed  PubMed Central  Google Scholar 

  5. Friston, K.J., Lawson, R. & Frith, C.D. On hyperpriors and hypopriors: comment on Pellicano and Burr. Trends Cogn. Sci. 17, 1 (2013).

    PubMed  Google Scholar 

  6. Lawson, R.P., Friston, K.J. & Rees, G. A more precise look at context in autism. Proc. Natl. Acad. Sci. USA 112, E5226 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. Palmer, C.J., Lawson, R.P. & Hohwy, J. Bayesian approaches to autism: towards volatility, action, and behavior. Psychol. Bull. 143, 521–542 (2017).

    PubMed  Google Scholar 

  8. Behrens, T.E., Woolrich, M.W., Walton, M.E. & Rushworth, M.F. Learning the value of information in an uncertain world. Nat. Neurosci. 10, 1214–1221 (2007).

    Article  CAS  PubMed  Google Scholar 

  9. Browning, M., Behrens, T.E., Jocham, G., O'Reilly, J.X. & Bishop, S.J. Anxious individuals have difficulty learning the causal statistics of aversive environments. Nat. Neurosci. 18, 590–596 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  10. Behrens, T.E.J., Hunt, L.T., Woolrich, M.W. & Rushworth, M.F.S. Associative learning of social value. Nature 456, 245–249 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  11. Mathys, C.D. et al. Uncertainty in perception and the Hierarchical Gaussian Filter. Front. Hum. Neurosci. 8, 825 (2014).

    PubMed  PubMed Central  Google Scholar 

  12. Iglesias, S. et al. Hierarchical prediction errors in midbrain and basal forebrain during sensory learning. Neuron 80, 519–530 (2013).

    CAS  PubMed  Google Scholar 

  13. Marshall, L. et al. Pharmacological fingerprints of contextual uncertainty. PLoS Biol. 14, e1002575 (2016).

    PubMed  PubMed Central  Google Scholar 

  14. de Berker, A.O. et al. Computations of uncertainty mediate acute stress responses in humans. Nat. Commun. 7, 10996 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. Yu, A. & Dayan, P. Expected and unexpected uncertainty: ACh and NE in the neocortex. Adv. Neural Inf. Process. Syst. 15, 173–180 (2003).

    Google Scholar 

  16. Berridge, C.W. & Waterhouse, B.D. The locus coeruleus-noradrenergic system: modulation of behavioral state and state-dependent cognitive processes. Brain Res. Brain Res. Rev. 42, 33–84 (2003).

    PubMed  Google Scholar 

  17. Hasselmo, M.E. & McGaughy, J. High acetylcholine levels set circuit dynamics for attention and encoding and low acetylcholine levels set dynamics for consolidation. Prog. Brain Res. 145, 207–231 (2004).

    CAS  PubMed  Google Scholar 

  18. Kobayashi, M. et al. Selective suppression of horizontal propagation in rat visual cortex by norepinephrine. Eur. J. Neurosci. 12, 264–272 (2000).

    CAS  PubMed  Google Scholar 

  19. Shepard, K.N., Liles, L.C., Weinshenker, D. & Liu, R.C. Norepinephrine is necessary for experience-dependent plasticity in the developing mouse auditory cortex. J. Neurosci. 35, 2432–2437 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  20. Lawson, R.P., Aylward, J., White, S. & Rees, G. A striking reduction of simple loudness adaptation in autism. Sci. Rep. 5, 16157 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  21. Ewbank, M.P. et al. Repetition suppression in ventral visual cortex is diminished as a function of increasing autistic traits. Cereb. Cortex 25, 3381–3393 (2015).

    PubMed  Google Scholar 

  22. Gomot, M. et al. Candidate electrophysiological endophenotypes of hyper-reactivity to change in autism. J. Autism Dev. Disord. 41, 705–714 (2011).

    PubMed  Google Scholar 

  23. Kleinhans, N.M. et al. Reduced neural habituation in the amygdala and social impairments in autism spectrum disorders. Am. J. Psychiatry 166, 467–475 (2009).

    PubMed  Google Scholar 

  24. den Ouden, H.E., Daunizeau, J., Roiser, J., Friston, K.J. & Stephan, K.E. Striatal prediction error modulates cortical coupling. J. Neurosci. 30, 3210–3219 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  25. Yu, A.J. Change is in the eye of the beholder. Nat. Neurosci. 15, 933–935 (2012).

    CAS  PubMed  Google Scholar 

  26. Solomon, M., Smith, A.C., Frank, M.J., Ly, S. & Carter, C.S. Probabilistic reinforcement learning in adults with autism spectrum disorders. Autism Res. 4, 109–120 (2011).

    PubMed  PubMed Central  Google Scholar 

  27. South, M., Newton, T. & Chamberlain, P.D. Delayed reversal learning and association with repetitive behavior in autism spectrum disorders. Autism Res. 5, 398–406 (2012).

    PubMed  Google Scholar 

  28. Nemeth, D. et al. Learning in autism: implicitly superb. PLoS One 5, e11731 (2010).

    PubMed  PubMed Central  Google Scholar 

  29. Brown, J., Aczel, B., Jiménez, L., Kaufman, S.B. & Grant, K.P. Intact implicit learning in autism spectrum conditions. Q. J. Exp. Psychol. (Hove) 63, 1789–1812 (2010).

    Google Scholar 

  30. Ratcliff, R., Smith, P.L., Brown, S.D. & McKoon, G. Diffusion decision model: current issues and history. Trends Cogn. Sci. 20, 260–281 (2016).

    PubMed  PubMed Central  Google Scholar 

  31. Wiecki, T.V., Sofer, I. & Frank, M.J. HDDM: hierarchical Bayesian estimation of the drift-diffusion model in Python. Front. Neuroinform. 7, 14 (2013).

    PubMed  PubMed Central  Google Scholar 

  32. Baron-Cohen, S., Wheelwright, S., Skinner, R., Martin, J. & Clubley, E. The autism-spectrum quotient (AQ): evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. J. Autism Dev. Disord. 31, 5–17 (2001).

    CAS  PubMed  Google Scholar 

  33. Mathys, C., Daunizeau, J., Friston, K.J. & Stephan, K.E. A bayesian foundation for individual learning under uncertainty. Front. Hum. Neurosci. 5, 39 (2011).

    PubMed  PubMed Central  Google Scholar 

  34. Lam, K.S., Aman, M.G. & Arnold, L.E. Neurochemical correlates of autistic disorder: a review of the literature. Res. Dev. Disabil. 27, 254–289 (2006).

    PubMed  Google Scholar 

  35. Daluwatte, C. et al. Atypical pupillary light reflex and heart rate variability in children with autism spectrum disorder. J. Autism Dev. Disord. 43, 1910–1925 (2013).

    PubMed  PubMed Central  Google Scholar 

  36. Courchesne, E., Kilman, B.A., Galambos, R. & Lincoln, A.J. Autism: processing of novel auditory information assessed by event-related brain potentials. Electroencephalogr. Clin. Neurophysiol. 59, 238–248 (1984).

    CAS  PubMed  Google Scholar 

  37. Jeste, S.S. et al. Electrophysiological evidence of heterogeneity in visual statistical learning in young children with ASD. Dev. Sci. 18, 90–105 (2015).

    PubMed  Google Scholar 

  38. Falck-Ytter, T. & von Hofsten, C. How special is social looking in ASD: a review. Prog. Brain Res. 189, 209–222 (2011).

    PubMed  Google Scholar 

  39. Aston-Jones, G. & Cohen, J.D.A.N. An integrative theory of locus coeruleus-norepinephrine function: adaptive gain and optimal performance. Annu. Rev. Neurosci. 28, 403–450 (2005).

    CAS  PubMed  Google Scholar 

  40. Costa, V.D. & Rudebeck, P.H. More than meets the eye: the relationship between pupil size and locus coeruleus activity. Neuron 89, 8–10 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. Hasselmo, M.E., Linster, C., Patil, M., Ma, D. & Cekic, M. Noradrenergic suppression of synaptic transmission may influence cortical signal-to-noise ratio. J. Neurophysiol. 77, 3326–3339 (1997).

    CAS  PubMed  Google Scholar 

  42. Hirata, A., Aguilar, J. & Castro-Alamancos, M.A. Noradrenergic activation amplifies bottom-up and top-down signal-to-noise ratios in sensory thalamus. J. Neurosci. 26, 4426–4436 (2006).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. Martins, A.R.O. & Froemke, R.C. Coordinated forms of noradrenergic plasticity in the locus coeruleus and primary auditory cortex. Nat. Neurosci. 18, 1483–1492 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  44. Balsters, J.H. et al. Disrupted prediction errors index social deficits in autism spectrum disorder. Brain 140, 235–246 (2017).

    PubMed  Google Scholar 

  45. Manning, C., Kilner, J., Neil, L., Karaminis, T. & Pellicano, E. Children on the autism spectrum update their behaviour in response to a volatile environment. Dev. Sci. http://dx.doi.org/10.1111/desc.12435 (2016).

  46. Happé, F.G. et al. Demographic and cognitive profile of individuals seeking a diagnosis of autism spectrum disorder in adulthood. J. Autism Dev. Disord. 46, 3469–3480 (2016).

    PubMed  Google Scholar 

  47. Haker, H., Schneebeli, M. & Stephan, K.E. Can Bayesian theories of autism spectrum disorder help improve clinical practice? Front. Psychiatry 7, 107 (2016).

    PubMed  PubMed Central  Google Scholar 

  48. Corlett, P.R. & Fletcher, P.C. Computational psychiatry: a Rosetta Stone linking the brain to mental illness. Lancet Psychiatry 1, 399–402 (2014).

    PubMed  Google Scholar 

  49. Teufel, C. & Fletcher, P.C. The promises and pitfalls of applying computational models to neurological and psychiatric disorders. Brain 139, 2600–2608 (2016).

    PubMed  PubMed Central  Google Scholar 

  50. Sevgi, M., Diaconescu, A.O., Tittgemeyer, M. & Schilbach, L. Social Bayes: using Bayesian modeling to study autistic trait–related differences in social cognition. Biol. Psychiatry 80, 112–119 (2016).

    PubMed  Google Scholar 

  51. American Psychiatric Association. Diagnostic and Statistical Manual–Text Revision (DSM-IV-TRim, 2000) (American Psychiatric Association, 2000).

  52. World Health Organization. International classification of diseases (ICD-10) (World Health Organization, 2012).

  53. Wechsler, D. & Hsiao-pin, C. WASI-II: Wechsler Abbreviated Scale of Intelligence (Pearson, 2011).

  54. Lord, C. et al. The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. J. Autism Dev. Disord. 30, 205–223 (2000).

    CAS  PubMed  Google Scholar 

  55. Woodbury-Smith, M.R., Robinson, J., Wheelwright, S. & Baron-Cohen, S. Screening adults for Asperger Syndrome using the AQ: a preliminary study of its diagnostic validity in clinical practice. J. Autism Dev. Disord. 35, 331–335 (2005).

    CAS  PubMed  Google Scholar 

  56. Ruzich, E. et al. Measuring autistic traits in the general population: a systematic review of the Autism-Spectrum Quotient (AQ) in a nonclinical population sample of 6,900 typical adult males and females. Mol. Autism 6, 2 (2015).

    PubMed  PubMed Central  Google Scholar 

  57. Willenbockel, V. et al. Controlling low-level image properties: the SHINE toolbox. Behav. Res. Methods 42, 671–684 (2010).

    PubMed  Google Scholar 

  58. Bonnet, C., Fauquet Ars, J. & Estaún Ferrer, S. Reaction times as a measure of uncertainty. Psicothema 20, 43–48 (2008).

    PubMed  Google Scholar 

  59. Vossel, S. et al. Spatial attention, precision, and Bayesian inference: a study of saccadic response speed. Cereb. Cortex 24, 1436–1450 (2014).

    PubMed  Google Scholar 

  60. Hosmer, D.W. Jr. & Lemeshow, S. Applied Logistic Regression (Wiley, 2004).

  61. Ioannidis, J.P. Why most discovered true associations are inflated. Epidemiology 19, 640–648 (2008).

    PubMed  Google Scholar 

  62. Jackson, I. & Sirois, S. Infant cognition: going full factorial with pupil dilation. Dev. Sci. 12, 670–679 (2009).

    PubMed  Google Scholar 

  63. Kang, O. & Wheatley, T. Pupil dilation patterns reflect the contents of consciousness. Conscious. Cogn. 35, 128–135 (2015).

    PubMed  Google Scholar 

  64. Knapen, T. et al. Cognitive and ocular factors jointly determine pupil responses under equiluminance. PLoS One 11, e0155574 (2016).

    PubMed  PubMed Central  Google Scholar 

  65. Schwarzkopf, D.S., Anderson, E.J., de Haas, B., White, S.J. & Rees, G. Larger extrastriate population receptive fields in autism spectrum disorders. J. Neurosci. 34, 2713–2724 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  66. Groppe, D.M., Urbach, T.P. & Kutas, M. Mass univariate analysis of event-related brain potentials/fields I: a critical tutorial review. Psychophysiology 48, 1711–1725 (2011).

    PubMed  PubMed Central  Google Scholar 

  67. Rescorla, R.A. & Wagner, A.R. A theory of Pavlovian conditioning: variations in the effectiveness of reinforcement. in Classical Conditioning II: Current Research and Theory 64–99 (Appleton-Century-Crofts, 1972).

  68. Rigoux, L., Stephan, K.E., Friston, K.J. & Daunizeau, J. Bayesian model selection for group studies - revisited. Neuroimage 84, 971–985 (2014).

    CAS  PubMed  Google Scholar 

  69. Kinnealey, M., Oliver, B. & Wilbarger, P. A phenomenological study of sensory defensiveness in adults. Am. J. Occup. Ther. 49, 444–451 (1995).

    CAS  PubMed  Google Scholar 

Download references

Acknowledgements

This work was supported by a Wellcome Trust Senior Clinical Research Fellowship (100227: G.R.). We thank all the participants who gave their time to take part in this research and M. Browning for helpful comments on an earlier poster presentation of these data.

Author information

Authors and Affiliations

  1. Wellcome Trust Centre for Neuroimaging, University College London, London, UK

    Rebecca P Lawson, Christoph Mathys & Geraint Rees

  2. Institute of Cognitive Neuroscience, University College London, London, UK

    Rebecca P Lawson & Geraint Rees

  3. Department of Psychology, University of Cambridge, Cambridge, UK

    Rebecca P Lawson

  4. Scuola Internazionale Superiore di Studi Avanzati (SISSA), Trieste, Italy

    Christoph Mathys

  5. Max Planck UCL Centre for Computational Psychiatry and Ageing, University College London, London, UK

    Christoph Mathys

  6. Translational Neuromodeling Unit, Institute for Biomedical Engineering, University of Zurich and Eidgenössische Technische Hochschule (ETH) Zurich, Zurich, Switzerland

    Christoph Mathys

Authors
  1. Rebecca P Lawson
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Christoph Mathys
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Geraint Rees
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

R.P.L. conceived the study and collected and analyzed the data. R.P.L. and C.M. modeled the data. R.P.L., C.M. and G.R. wrote the manuscript.

Corresponding author

Correspondence to Rebecca P Lawson.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Reaction time as a function of stimulus noise.

Collapsing across the three levels of expectedness, the non-significant stimulus noise*group interaction indicates that the linear relationship between noise and RT was equivalent in both groups (ASD, n=24; NT, n=25). See main text for supporting statistics. ASD, autism spectrum disorder. NT, neurotypical. RT, reaction time. Data points represent individual participants, red lines indicate the mean, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean for each condition and group. H=high noise, M=medium noise and N=no noise

Supplementary Figure 2 Inverse efficiency scores.

To ensure that the attenuated UE-E RT difference in the ASD participants was robust to correction accuracy, we calculated inverse efficiency scores (IES) as RT/(1-accuracy) for each condition. As for the analysis of RT and error rates alone, see main manuscript, there was a significant main effect of expectedness (F(1.8,83.11)=34.24, P<0.001) and noise (F(2,94)=6.87, P=0.002) and again only the expectedness*group interaction was significant in this analysis (F(2,94)=9.98, P<0.001). The noise*group (F(2,94)=0.24, P=0.79) and expectedness*noise*group interactions were not significant (F(4,188)=2.2, P=0.07). Thus, our primary reaction time finding is robust to correction condition-specific accuracy (ASD, n=24; NT, n=25). ASD, autism spectrum disorder. NT, neurotypical. RT, reaction time. Data points represent individual participants, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean for each condition and group. E=expected, N=neutral and UE=unexpected.

Supplementary Figure 3 Caution of responding control analysis.

To exclude the possibility that our group difference in UE-E RT (i.e. reduced behavioural surprise in ASD) is explainable by increased response caution in the ASD participants we compared the 12 fastest responders from the ASD group (mean RT 418 ms) against the 12 slowest responders in the NT group (mean RT 540 ms) on the primary UE-E RT difference measure. Here the 12 fastest overall responding ASD participants are those who are most impulsive/least cautious in general responding (i.e. have the lowest response thresholds) whereas the 12 slowest overall NTs are the least impulsive/most cautious (i.e. have the highest response thresholds). Indeed, mean reaction time is significantly faster in this subgroup of ASD participants than in the subgroup of NTs (t(22)=3.38, P=0.03). Nonetheless, independent-samples t-tests revealed that the ASD participants (in this subset of fast general responders) still show significantly diminished behavioural surprise (t(22)=2.39, P=0.026) relative to NTs (in this subset of slow responders). ASD, autism spectrum disorder. NT, neurotypical. RT, reaction time. Data points represent individual participants, red lines indicate the mean, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean. Star indicates significance at P<0.05.

Supplementary Figure 4 Replication of behavioral results in a nonclinical sample.

(a) The same task conducted in a sample of non-clinical volunteers characterised according to high or low autistic traits (AQ) replicates the interaction between expectedness (E=expected, N=neutral, UE=unexpected) and autistic tendency (high AQ, n=26; low AQ, n=31). There was a significant main effect of expectedness (F(2,110)=69.46, P<0.001) and, crucially, a significant expectedness*AQ group interaction (F(2,110)=13.29, P<0.001); suggesting that participants with high AQ scores show a reduced modulation of RT as a function of expectedness (e.g. reduced slope), relative to participants with low AQ scores. There was a main effect of noise (F(2,110)=16.96, P<0.001), and noise*group interaction (F(2,110)=5.07, P=0.008). No other linear interactions or main effects were significant (P’s>0.2). (b) An independent samples t-test demonstrated that behavioural surprise was significantly attenuated in the high AQ group (t(55)=4.32, P<0.001). (c, d) Error rates were subject to the same analysis as above. There was a significant main effect of expectedness (F(2,110)=19.89, P<0.001) but the expectedness*AQ group interaction did not reach significance (F(2,110)=.85, P=0.42); suggesting that the main effect of expectedness on accuracy did not vary as a function of autistic traits. The main effect of noise narrowly missed significance (F(2,110)=2.36, p=0.09), but there was no noise*group interaction (F(2,110)=1.67, P=0.1). One low AQ participant showed relatively high % errors in the UE condition, but their overall errors were within reasonable limits and results are not changed if they are excluded. Compare with Figure 2a-d in the main text. Data points represent individual participants, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean. Star indicates significance at P<0.05

Supplementary Figure 5 Responses to face and house stimuli.

To confirm that there were no group differences in RTs or error rates in responding to the different outcome image types (faces, houses) we examined these responses in two separate repeated-measures ANOVAs with group (ASD, n=24; NT, n=25) as a between participants factor in each case. For reaction times there was a significant main effect of stimulus type, reflecting the fact that participants were in general slower to respond to house images over face images (F(1,47)=16.52, P<0.001). Additionally there was a main effect of group indicating that the ASD participants were generally slower to respond than the NT participants (F(1,47)=5.54, P=0.023) but crucially there was no interaction between stimulus type and group (F(1,47)=1.23, P=0.2). For error rates, participants generally made more errors on house trials (main effect of stimulus type: F(1,47)=13.37, P=0.001), but there was no group difference in errors overall (non-significant main effect of group: F(1,47)=0.8, P=0.37) and there was no stimulus type x group interaction (F(1,47)=0.02, P=0.9). ASD, autism spectrum disorder. NT, neurotypical. Data points represent individual participants, red lines indicate the mean, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean.

Supplementary Figure 6 Results of Bayesian-model selection.

The protected exceedance probability from the Bayesian Model Selection (BMS) of log model evidences shows that the 3-level HGF (HGF-3) describes subject’s behaviour better than alternative learning models (RW; Rescorla Wagner, SK1; Sutton K1, HGF-2; 2-level Hierarchical Gaussian Filter). See main text for details.

Supplementary Figure 7 Model-simulated reaction times.

As an additional validation of the HGF model performance we simulated trial-wise RTs using the fitted perceptual and response model parameters from each of our 24 ASD and 25 NT participants. These simulations can recover the group differences in the main behavioural effect of expectation (compare to Figure 2a&b in the main manuscript). Statistical analysis of these model simulated RTs indicates a significant expectedness * group interaction (F(1,94)=4.44, P=0.014), and the simulated UE-E RT difference was significantly lower when simulated from the ASD parameters, relative to the NT parameters(t(47)=2.57, P=0.013). ASD, autism spectrum disorder. NT, neurotypical. UE, unexpected. N, non-predictive. E, expected. Data points represent the mean of 32 simulations for each individual participant, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the group mean, red lines indicate the group mean. Star indicates significance at P<0.05

Supplementary Figure 8 Average HGF-parameter estimates across groups.

Individual participant parameter estimates for each of the free parameters estimated from the HGF, for both the ASD (n=24) and NT (n=25) groups. A statistically significant MANOVA effect indicated that the groups would differ on one or more of the estimated model parameters, Pillai’s’ Trace =.43, F(8, 40) = 3.81, P=0.002). Independent samples t-tests indicate a significant group difference in baseline log RT (β0; t(47) = 2.33, P=0.024), phasic volatility (β4; t(47) = 2.15, P=0.037) and tonic volatility at the third level (ω3; t(47)=2.10, P=0.045). Outcome surprise (β1; t(47) = -1.73, P=0.09) and outcome uncertainty (β2; t(47) = -1.87, P=0.06) narrowly missed significance. There were no group differences in probability uncertainty (β3; t(47) = -.51, P=0.61), decision noise (ζ; t(47) = -.55, P=0.59) or tonic volatility at the second level (ω2; -.21, P=0.84). See the main text and Figure 4b for a multiple linear regression analysis predicting group status from these same parameters. ASD, autism spectrum disorder. NT, neurotypical. Data points represent individual participants, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean, red lines indicate the group mean. Star indicates significance at P<0.05

Supplementary Figure 9 Pupil size and dynamic learning rates.

The analysis reported in the main text indicates a sustained positive relationship between pupil size and precision-weighted prediction errors (ɛ3) in the ASD participants (Figure 5b). The precision weight (on the prediction error) is proportional to the update of environmental volatility and is formally related to dynamic trial-wise learning rate (α3) This additional analysis indicates that the learning rates themselves (α2 and α3) do not have a significant influence on pupil dilation in either group. As for the results reported in the main text (see Online Methods) this regression analysis included, trial type (face, house), fixation compliance, mean RT and UE-E ground truth contrasts, as control regressors. Shaded regions represent standard error of the mean.

Supplementary Figure 10 Pupil size and precision-weighted PEs in stable and volatile task periods.

The pupil regression reported in the main text (Figure 5b) examined the relationship between precision-weighted prediction errors (ɛ3; PE’s) and pupil size across all trials in the experiment. A strength of this analysis is that it represents the pupil response when each participant was actually surprised, and does not impose knowledge of the task structure. Nonetheless, to examine the relationship between precision-weighted PE and pupil size in the volatile and stable periods of the task we conducted the same regression analysis (see main text and online methods) but separately for the 72 ‘stable’ trials and ‘72’ volatile trials (see Figure 1) towards the end of the experiment. (left) In the stable period there is no relationship between precision-weighted prediction errors and pupil size in either group or no differences between the groups. (right) The relationship between precision-weighted PE’s and pupil size in the ASD participants (blue) is apparent 1000ms after the outcome appears in the volatile period of the task. Blue solid line shows where the ASD participants differ from zero and black dotted line shows where the ASD participants differed from the NT participants. Shaded region represents standard error of the mean. Consistent with the analysis of learning rates in the volatile and stable task periods (Figure 3c), this suggests that the ASD participants tend to show aberrant noradrenergic surprise about volatility, in response to volatility (e.g. over-updating learning about volatility and over-engaging noradrenergic responses to surprise about volatility, in the face of environmental volatility). However, we caution against the low trial numbers included in this analysis (72, vs a maximum of 456 in the analysis reported in the main text) and the fact that one control participant did not have enough good trials in the volatile period to be included in this analysis, so participant numbers are also reduced (ASD=11, NT=13).

Supplementary Figure 11 Fixation compliance across trial types.

Mean absolute deviation (MAD) from fixation (in degrees of visual angle) across groups and conditions. Generally fixation compliance was very good, <1° of visual angle in both the vertical and horizontal axes. Stimulus duration was purposefully short to eliminate saccades. Crucially there is no systematic difference in fixation compliance that would impact on the pupillometry results, either across groups or conditions. One participant showed systematically larger deviation from fixation in the horizontal plane, though fixation was still good (below 2° visual angle) and not beyond the physical limits of the stimulus being presented. Importantly, trial-wise absolute deviation from fixation was included as a regressor of no interest in all pupillometry analyses reported in the main text and supplemental results, and thus our pupillometry analyses are corrected for eye movements. ASD, autism spectrum disorder. NT, neurotypical. UE, unexpected. N, non-predictive. E, expected. Data points represent individual participants, shaded regions and error bars show 95% confidence intervals and 1 standard deviation of the mean, red lines indicate the group mean.

Supplementary Figure 12 Pupil size and reaction time.

The relationship between precision-weighted prediction errors and pupil size reported in the main text (Figure 5b), links pupil size to behaviour indirectly via the HGF model, since the precision-weighted prediction errors are estimated for each participant on the basis of their trial-wise RT. However, we conducted an additional regression analysis to investigate the relationship between basic behaviour (trial-wise RT) and pupil size directly. As RT increases, post-outcome pupil size shows an initial decrease from baseline followed by an increase towards the end of the trial. Crucially, there are no time points in which the relationship between RT and pupil size is significantly different between the ASD and NT groups. Notably, trial-wise RT is included as a control regressor in the results reported in the main text (Figure 5), and in the analyses reported above (Figure S9 & S10), so where there is a significant relationship between pupil size and the trial-wise model parameters this exists over and above any effect of RT on pupil size. Blue and yellow bars indicates where the relationship between RT and pupil size significantly differed from zero in the ASD and NT participants, respectively. As for the results reported in the main text (see online methods) this regression analysis included, trial type (face, house), fixation compliance and UE-E ground truth contrasts, as control regressors. Shaded region represents standard error of the mean.

Supplementary Figure 13 Raw pupil traces.

(a) Raw mean pupil dilation in ASD participants (blue) and NT participants (yellow) separated into trials in which the outcome was unexpected (UE: dotted line) and trials where the outcome was expected (E: solid line) (a) The UE-E difference (i.e. ground truth ‘surprise’) in the ASD participants (blue) and NT participants (yellow). Equivalent to the UE-E contrast from the regression model presented in Figure 5a. Shaded region shows standard error of the mean.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–13 and Supplementary Tables 1 and 2 (PDF 1509 kb)

Life Sciences Reporting Summary (PDF 110 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Lawson, R., Mathys, C. & Rees, G. Adults with autism overestimate the volatility of the sensory environment. Nat Neurosci 20, 1293–1299 (2017). https://doi.org/10.1038/nn.4615

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nn.4615

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing