July 24, 2020

Seeing the Future: A Predictive Coding Explanation of the Brain

By Kennedy Salamat

How exactly does the brain process information? Does one’s brain observe the world around it and subsequently react, or does it already have an idea of what is coming next that it updates through new information (1)? Predictive coding encompasses the latter idea, hypothesizing that the brain predicts what is about to happen and modifies its understanding based on sensory information (2). An analogy for predictive coding is baking cookies. After taking the pan out of the oven, would you immediately grab a cookie and jerk your hand back because the pan had burned it, or would you wait a few minutes, tentatively start to pick up a cookie, and eat it, only if it had cooled down? The process of making a prediction about the temperature of the cookie and then using sensory information to revise one’s assessment as to whether it had cooled down parallels predictive coding theory.

The idea of predictive coding is increasingly being used by scientists to explain neurological disorders (3). When the balance between predictions and sensory information is thrown off, individuals may not be able to adapt to a new environment or make sense of everyday situations. For example, a study linked Autism Spectrum Disorder (ASD) to instances where an individual’s brain considers its own predictions more heavily than the sensory information it receives (3). This may cause these individuals to resist change, struggle to understand a conversation, or be unable to formulate a response in day-to-day interactions (4) . Moreover, predictive coding is used by scientists to try to explain depression and other mental health conditions. Depression is thought to occur when individuals’ brains heavily weigh their own predictions -- which are typically that the situation is negative -- and are unable to adapt their views when presented with something positive (5). Another example is Attention Deficit Hyperactivity Disorder (ADHD). In this case, some scientists believe that the brain more heavily considers sensory information as opposed to its prediction, causing individuals to become easily distracted (3). By utilizing the predictive coding model as a lens, one may gain deeper insight into various psychological phenomena, such as mania, depression, bipolar disorder, ASD, ADHD, and schizophrenia, thus generating more effective treatment options (6).

For example, researchers used predictive coding to understand forced normalization in epilepsy patients after surgery, which is defined as “the emergence of psychoses following the establishment of seizure control” (7). By using predictive coding as a lens, the researchers concluded one way to mitigate this would be to have a “preoperative intervention addressing patient predictions,” enabling the patient to start to process this new change in their life, and allow their brain to adapt its predictions (8). Researchers also believe that predictive coding may be useful in shaping psychosis diagnoses and treatments. By understanding the various types of psychosis through a predictive coding framework, researchers may be able to better distinguish between disorders, more effectively use medications, and improve therapy treatments (6).

In order to be able to fully understand disorders through the predictive coding framework, computational models are needed (6). Recently, a team of researchers from the University of Geneva created a computer model of how the brain processes speech using predictive coding techniques and brainwave information (9). The system read in sound waves and detected where in the sound wave a syllable started and ended (10). The system then combined computer generated predictions and information from the sound wave to arrive at the correct syllable (10). The model successfully classified 2,888 syllables during the testing phase (9).  Building off of the University of Geneva researchers’ computational model, perhaps future models may alter how predictions and sensory information are weighted with the purpose of precisely understanding how information is processed in the aforementioned disorders. This would help verify theories which rely on predictive coding to explain neurological disorders, ultimately increasing understanding of these disorders and enabling the design of more effective medical solutions. 

 


References:

  1. Briggs, S. (2019). How Predictive Coding Is Changing Our Understanding of the Brain. Retrieved from https://www.opencolleges.edu.au/informed/features/predictive-coding/
  2. de-Wit, L., Machilsen, B., & Putzeys, T. (2010). Predictive coding and the neural response to predictable stimuli. The Journal of neuroscience : the official journal of the Society for Neuroscience, 30(26), 8702–8703. https://doi.org/10.1523/JNEUROSCI.2248-10.2010
  3. Gonzalez-Gadea, M. L., Chennu, S., Bekinschtein, T. A., Rattazzi, A., Beraudi, A., Tripicchio, P., Moyano, B., Soffita, Y., Steinberg, L., Adolfi, F., Sigman, M., Marino, J., Manes, F., & Ibanez, A. (2015). Predictive coding in autism spectrum disorder and attention deficit hyperactivity disorder. Journal of neurophysiology, 114(5), 2625–2636. https://doi.org/10.1152/jn.00543.2015
  4. Musser, G. (2020, April 16). Predictive coding theory of autism, explained. Retrieved July 5, 2020, from https://www.spectrumnews.org/news/predictive-coding-theory-autism-explained/
  5. Clark, J. E., Watson, S., & Friston, K. J. (2018). What is mood? A computational perspective. Psychological medicine, 48(14), 2277–2284. https://doi.org/10.1017/S0033291718000430
  6. Sterzer, P., Adams, R. A., Fletcher, P., Frith, C., Lawrie, S. M., Muckli, L., Petrovic, P., Uhlhaas, P., Voss, M., & Corlett, P. R. (2018). The Predictive Coding Account of Psychosis. Biological psychiatry, 84(9), 634–643. https://doi.org/10.1016/j.biopsych.2018.05.015
  7. Loganathan, M. A., Enja, M., & Lippmann, S. (2015). FORCED NORMALIZATION: Epilepsy and Psychosis Interaction. Innovations in clinical neuroscience, 12(5-6), 38–41.
  8. Mehmood, S., Dale, C., Parry, M., Snead, C., & Valiante, T. A. (2017). Predictive coding: A contemporary view on the burden of normality and forced normalization in individuals undergoing epilepsy surgery. Epilepsy & behavior : E&B, 75, 110–113. https://doi.org/10.1016/j.yebeh.2017.06.042
  9. Université de Genève. (2020, June 26). Computational model decodes speech by predicting it. ScienceDaily. Retrieved July 2, 2020 from www.sciencedaily.com/releases/2020/06/200626114808.htm
  10. Hovsepyan, S., Olasagasti, I., & Giraud, A. L. (2020). Combining predictive coding and neural oscillations enables online syllable recognition in natural speech. Nature communications, 11(1), 3117. https://doi.org/10.1038/s41467-020-16956-5