Grant title: Neural mechanisms of prediction signalling along the cortical auditory pathway
Principal (co-)investigators: Jan Schnupp, Ryszard Auksztulewicz, Nicol Harper (Oxford)
Grant scheme: General Research Fund grant Nr 11100518 (9042615)
Funding amount: HK$ 896,913
Grant duration: 01.01.2019 - 31.12.2021
Abstract: One increasingly influential idea in neuroscience proposes that one of the most important tasks for the brain is not merely to process incoming sights and sounds, but to predict what will happen next. The brain generates “internal models” of the world that attempt to predict how the imminent future will unfold, and the predictions of the model are constantly compared against new sensory information, and errors in the predictions are flagged to keep the predictions on track with reality and to improve the internal model. Although there is some evidence for this type of “predictive coding”, for example, from bulk brain signals recorded through electrodes on the scalp, as yet there is little examination in single neurons (brain cells). This project will examine the prediction of future sounds which form parts of rhythmic auditory patterns by single neurons recorded in the auditory brains of rats. We will use carefully constructed rhythmic sequences, where, a small, randomly chosen proportion of the sounds deviate from the regular pattern by unpredictable jumps in pitch, duration, loudness or sound source position. Using surface and depth array electrodes we will then identify the location and characteristics of neurons in the animals’ cerebral cortex which register unexpected deviations along each of these perceptual attributes of sound. Predictive coding is now widely thought to be a key principle of brain function, which, if fully understood, could radically improve our capacity to ameliorate sensory deficits. Predictive coding is believed to involve feedback from higher brain areas to lower areas. Some evidence suggests that the auditory system uses predictive adaptation to better encode sounds in noisy environments. Unlike the ear, the action of hearing aids or cochlear implants (remarkable electronic ears that can be surgically implanted to allow hearing in the deaf) does not currently depend on brain activity. With an understanding of predictive neural phenomena, models of these phenomena could be included in the algorithms of hearing prosthetics.