Skip to main content
European Commission logo print header

Cellular mechanisms of predictive processing and its effects on perception

Periodic Reporting for period 1 - MECHPREDPRO (Cellular mechanisms of predictive processing and its effects on perception)

Reporting period: 2018-08-01 to 2020-07-31

Our brains consist of different types of neurons that together create our mental world. Layer 5 pyramidal (L5p) neurons are the key neurons of the cerebral cortex that integrate information across cortical layers and provide the main output of the cortex to other brain areas. In our project we tried to better understand the role of L5p in mental processes, in particular in perception.

It is easy to assume that perception of the world accurately reflects reality. However, work in psychology and neuroscience has demonstrated that perception is always a mixture of prior knowledge and sensory data. Hence, sometimes we do not perceive “what is out there”, but rather what the prior knowledge dictates. In fascinating examples, expectations can create perception of objects that are actually not there in the real world. Such illusory perception is not only pathological, but can also happen in healthy humans, for example if we expect something to happen or the context suggests that some object should be there. Another example of illusory perception are dreams. In our project we studied the neural mechanisms of illusory perception.

We also wanted to demonstrate how taking into account the biophysical complexities of L5p cells help us to understand perception, prediction and conscious experiences. It is often assumed that neurons can be captured by simply assuming one process that sums the input and provides the output. However, in our project we tried to show that L5p cells consist of two compartments that have distinctly different roles. Our mental processing depends on the integration of these two compartments.
We have studied the cellular mechanisms of perception and how the different compartments of the pyramidal cells contribute to it. In particular we investigated whether illusory percepts can be triggered from the apical dendrites of L5p cells. Apical dendrites are targeted by internal information, thus we expected that apical dendrites mediate perception that is dictated by prior knowledge. We have designed novel experiments to study the role of predictions in perception and the neural basis of illusory perception. We have also developed new technologies to study the interaction between the different compartments of L5p cells.

Based on these findings we have proposed a theory explaining how perception arises in the brain from the integrative properties of the dendrites of L5p cells (Aru et al., 2020 in Trends in Cognitive Sciences). Conscious experiences are dependent on the interaction between the different compartments of pyramidal cells, modulated by subcortical structures like the thalamus. The fact that apical dendrites - carrying internal information - have a crucial influence on conscious experiences is in line with research showing that conscious experience is modulated by predictions and prior knowledge. We have also proposed how these fundamental cellular properties of internally generated perception can explain phenomena like dreams and illusory perception (Aru et al., 2020 in Neuroscience & Biobehavioral Reviews). During internally generated perception, input from the internal generative model targets the apical dendrites and can under certain circumstances dictate perception. This process is controlled by neuromodulation (Aru et al., 2020 in Neuroscience & Biobehavioral Reviews).
We have summarized the main contributions of the project as a novel neurobiological theory of consciousness (Aru et al., 2020a), which hopefully will provide useful insights for several fields. For example, for cognitive science our work provides a concrete neurobiological foundation for understanding the effects of prior knowledge and expectations on perception. Our theory also helps to better understand the potential neurobiological implementations of the predictive processing framework. In neuroscience, our work suggests many new experiments aiming to understand internally generated perception. For artificial intelligence our results provide a biological basis for understanding the success of attention mechanisms for machine translation and deep learning.

Our paper on the neurobiological mechanisms of internally driven perception (Aru et al., 2020b) will be useful for researchers from various fields. First, here we proposed a concrete neural mechanism of hallucinations, which will be of interest for scholars working on schizophrenia. Second, the neural mechanisms are amenable to testing by the neuroscience community. Third, as the paper also discusses the effects of neuromodulation on L5p cells and internally generated perception, the work will guide further experiments on the field of neuromodulatory effects on cognition. Finally, the paper will be of interest for researchers working on sleep and dreaming.

Taken together, we have shown that it is essential to take into account that cortical pyramidal cells have two distinct parts. Through their interactions we can better understand central phenomena of our mental worlds. This project, through studying the neural mechanisms of conscious experience, predictions and illusory perception advances our understanding of the neural mechanisms that underlie our mental lives.
Our project shows that the cellular details of pyramidal cells help to understand mental processes