Consciousness & Cognition


  • We focused on the multiscale understanding of different brain states: physiological (e.g. sleep, awake), drug-induced (e.g. anaesthesia) and pathological (e.g. disorders of consciousness). The study of the mechanisms and spatiotemporal properties of brain states provide the building blocks that support consciousness and cognition.

  • A main objective of this focus area was to collaboratively generate integrated data and computational models supporting the understanding of different brain-states, transitions across them, network complexity and cognitive functions.

  • In parallel, we developed an ethical and philosophical framework for the experimental and computational explorations of cognition and consciousness. 


Complex to abstract: the fascinating tree structure of dendrites can now be modelled at many scales. © eLife

Collaborative multiscale approach

From the structural point of view, the brain comprises highly intricate networks between neurons, nuclei, and areas. From the functional perspective, the brain can express a large variety of dynamical patterns, ranging from highly synchronous (sleep, deep anaesthesia, epilepsy) to asynchronous states (awake, attention). There are other conditions in which different functional states are expressed, as in the presence of drugs (e.g., anaesthesia), or in pathological conditions (e.g., disorders of consciousness, stroke, epilepsy). In this focus area, we aim at the multiscale understanding of different brain states and how they can support consciousness and cognition. It is also our objective to obtain a better, mechanism-based quantification of consciousness levels and contents, as well as new approaches to the treatment of alterations of consciousness following brain damage. To achieve these objectives, we have collaborated with 40 research groups, comprising a large variety of experimental, clinical, analytical, and computational methodologies that aimed at advancing our understanding of how cognition and consciousness emerge from brain networks. We worked hand in hand with philosophers and ethicists to develop a philosophical and ethical framework for the experimental and computational explorations of cognition and consciousness.

Showcase 3. Brain Complexity and Consciousness  demonstrated the simulation of unconscious states in three species: human, mouse, and monkey. The generation of unconscious states, like the effect of anaesthesia or the sleep-wake cycle, could be related to changes in microscopic parameters, such as the level of hyperpolarisation of the neurons, changes in GABA-A-receptor kinetics, or augmentation of spike-frequency adaptation. In Showcase 3, both empirically observed spontaneous and evoked dynamics could be simulated using anatomical and connectivity data from all three species. Moreover, Showcase 3 also focused on the effects of anaesthesia and the responsiveness of cortical networks at a large-scale level using TVB, with a spatial resolution that approaches the one of a cortical column. The fine-tuning of the microscopic parameters enables the study of a wide range of conditions, thus allowing the study of the universality of slow waves between species and between the different mechanisms underlying unconscious states.

Simulation of brain states in mouse, macaque and human. Whole-brain simulations of wake-like (conscious) and anaesthesia (unconscious) states in three different species. (References: Goldman et al. Biorxiv 2020Goldman et al. Biorxiv 2021).

Find out more about the Showcase 3 here.

Watch Showcase 3 video


Showcase 4 Object Perception and Memory built predictive coding models combining cognitive and perceptual mechanisms with the dynamics and neuronal circuitry characteristic of the cortex. To illustrate this process, an integrative suite of three complementary models with different levels of biological realism has been developed: 1) a neurobiological model of spiking cortical columns, 2) a cognitive model of invariant object perception and 3) a hybrid model of a predictive coding based spiking neural network. Showcase 4 presents the highlights of each of these models and elaborates how the knowledge gained from these models can be integrated to improve our understanding of perception and brain diseases.

Figure reference: A robot on EBRAINS has learned to combine vision and touch - EBRAINS. (n.d.).  retrieved June 28, 2022, from

Find out more about showcase 4 here.

Check out Showcase 4 latest video.


Cognition: Object recognition

To enable perception and recognition of objects through multiple senses, the brain must segregate (separately identify) and integrate information about their properties. We do not yet understand how these fundamental processes are mediated by neural mechanisms, especially in the neocortex. We investigated critical questions that remain open: how does the brain construct the kind of representations of objects we become aware of? How is it that, in perception, we have the experience of a specific view on an object, but also recognize its identity despite variations in position or viewing angle? And how are cortical circuits involved in object representations? Which cortical laminae and cell types are involved? How can local and long-range circuit mechanisms involving specific cell types with dendritic interactions shape multisensory integration? How can computational mechanisms be scaled up to comprise larger networks, dynamically performing perceptual and cognitive operations? What role do motor systems and the principle of active sensing play in multisensory segregation and integration? By means of experimental models, data analytics, computational models, multi-scale simulations and research in neurorobotics, now available in EBRAINS, we were implementing object recognition in an artificial, multisensory agent that used this cognitive capability to navigate through environments.

Figure references:

A robot on EBRAINS has learned to combine vision and touch - EBRAINS. (n.d.).  retrieved June 28, 2022, from

Dora, S., Bohte, S. M., & Pennartz, C. M. A. (2021). Deep Gated Hebbian Predictive Coding Accounts for Emergence of Complex Neural Response Properties Along the Visual Cortical Hierarchy. ORIGINAL RESEARCH article Front. Comput. Neurosci., 28 July 2021   


People and Groups

Cyriel Pennartz - Department of Cognitive and Systems Neuroscience, University of Amsterdam,  FNWI, Swammerdam Institute for Life Sciences - NL
Giovanni Pezzulo - National Research Council of Italy, Institute of Cognitive Sciences and Technologies (ISTC-CNR), in Rome - IT
Markus Diesmann - Forschungszentrum Jülich, Institut für Neurowissenschaften und Medizin (INM) - DE
Sacha van Albada - Forschungszentrum Jülich, Institut für Neurowissenschaften und Medizin (INM) - DE
Hans Ekkehard Plesser - Norges miljo-og biovitenskaplige universitet, Fakultet for realfag og teknologi, Institutt for datavitenskap - NO
Walter Senn - Universitaet Bern, Institut für Physiologie, Computational Neuroscience lab - CH
Pieter Roelfsema - Netherlands Institute for Neuroscience - NL
Jorge Mejias - Universiteit van Amsterdam, Faculty of Science, Swammerdam Institute for Life Sciences - NL
Emrah Duzel - Deutsches Zentrum fuer Neurodegenerative Erkrankungen EV, Clinical Neurophysiology and Memory - DE
Matthew Larkum - Humboldt-Universität zu Berlin - DE
Lars Muckli - Centre for Cognitive Neuroimaging, School of Psychology & Neuroscience, University of Glasgow - UK

Quantification and modelling of different brain states

The cerebral cortex network can express a large variety of dynamical patterns, ranging from highly synchronous (sleep, deep anaesthesia, epilepsy) to asynchronous states (awake, attention). Each of these states are associated to different spatiotemporal patterns of activity, network complexity, information processing properties, behaviours and consciousness levels. How can the same network express all those different states and transitions across them? What are the underlying mechanisms? To identify, characterise and understand the network dynamics in such states, we explored the spontaeous activity emerging from the network across the different brain states, with a focus in the transitions across them, and we carry out a multi-scale comparison across levels, ranging from simple in vitro systems to clinical pathological states (disorders of consciousness, stroke) to capture the unique, common properties across them. Data-driven models will provided a unifying framework to single out an effective and reductionistic representation, capable ofreproducing in the same system the dynamical regimes observed at different scales. Such an approach gives access to a larger exploration of variables than those experimentally testable, generating predictions and open questions that then can be experimentally tested.