Centre for Neural Computation (CNC)

The vision of CNC is to understand the neural basis of high-level computation in cortical systems of the brain. For 2022, we wish to emphasize the following discoveries and developments: The Moser group has introduced and developed methods for high-resolution neural population recordings. With these methods they have shown that grid cells operate on a manifold with the topology of a torus, confirming a major prediction of continuous attractor network models. They have developed a miniature 2-photon microscope allowing activity to be monitored during unconstrained behavior across thousands of cells in multiple planes of dense neuronal populations. These developments transform the study of neural coding by providing access to dynamics expressed only in the joint activity of large neural populations. The Kentros group generates highly specific molecular genetic tools and uses them to study neural circuitries underlying memory. They investigated the specificity of direct entorhinal inputs to the hippocampus in neural network oscillations. They also showed that the pathological interactions between Amyloid-beta and Tau in Alzheimer’s disease (AD) likely occur in the stellate cells of the entorhinal cortex (EC). Since these are the first neurons to die in AD patients, this may provide novel avenues for cell-specific therapies for AD. The Witter group finalized two studies on networks connecting the two main EC subdivisions (LEC and MEC) with each other and with the hippocampus. They showed that the LEC and MEC are reciprocally connected, and MEC modulates LEC-hippocampal projections by way of long-range GABAergic and glutamatergic projections. The ventral hippocampus provides an output to all levels of MEC, likely modulating the hippocampal flow of information to the rest of the brain. Both findings urge for a critical revision of the functional concepts on the hippocampal memory system. The Whitlock group studies how sensory cortices in freely moving rodents combine sensory signals with 3D posture and movement. They found how distinct sensory cortices differentially employ posture and movement signals to support sensory encoding and discovered neural pathways by which visual signals reach the motor system, enabling visually-guided motor behavior. They also developed an open-source anatomical reconstruction software toolkit. The Navarro Schröder group study patients with transient global amnesia. They spearheaded a multi-disciplinary project to acquire and use advanced MRI equipment at the Norwegian 7T center, including elastography to measure tissue stiffness, a coil to measure the spinal cord, and a phosphorus spectroscopy coil to quantify metabolic changes in the brain. They also acquired new insights into how the brain represents the location of others. The Nigro group aims to unveil the computations underlying multisensory objects in the cortex and the role of different cell types. New findings describe sensory responses in perirhinal cortex. They discovered a new GABAergic population in association cortices and developed a viral tool to label and manipulate it. The Quattrocolo group studies how different cell types contribute to the maturation of neuronal circuits of the hippocampal region, how synapses change during development, and how specific genes or cell types (e.g., Cajal-Retzius cells) influence their maturation. The group integrates cutting edge-technique, such as transcriptomic and proteomic analysis, behavioral experiments, cell-type specific viral vector and transgenic mouse lines, and anatomical tracing. The Roudi group studies the role of single-neuron input-output functions in neural networks and recently extended the approach to neural network models of probabilistic learning. They have further developed methods for analysing neural data and comparing neural codes using featureless inference where assumptions about what is coded is not a prerequisite. A new avenue is understanding how evolutionary dynamics in interacting agents can lead to the emergence of higher computational power. The Yaksi group studies how internal states of the brain interacts with sensory information and how these interactions are altered in health and disease. They have previously discovered how glia cells play a role in regulating excitability associated with seizures, and that distinct functional components of habenula, a brain region important for learning, are born in distinct developmental periods. The Ziaei group investigates neural mechanisms underlying emotional processes in aging. They have shown how aging impacts structural and functional brain circuitries involved during decoding and responding to social-emotional cues, which are critical abilities for effective social communication. An ongoing collaboration with St.Olav’s Hospital focuses on how these skills change with mood disorders in late adulthood. This will open doors for new psychotherapeutic treatments for patients with psychosocial difficulties.

The financial structure of the SFF has been revised several times during the 10 years period whereas several financial advisors has been involved in the reporting processes as well as the multiple revisions and amendments executed. Prior to 2020, costs associated with external funded projects with a scientific relationship with the SFF was included in the figures above. However, from 2021 onwards only costs related to the centre was reported, whereas the total activity including both the SFF as well as other external funded projected related to the centre were reported in the attached overview under section "special reports". The decision to revise the structure from 2021 onwards was made in close collaboration with the NFR, whilst it was proposed to revise the figures starting from 2013, it was decided to only use the structure from 2021 onwards, leaving the historic figures as previously reported. This is the reason for the major drop in in costs reported from 2020 to 2021. For the last two years the personnel costs are only reflecting the payroll of the personnel paid directly by the SFF funds and the total costs for the last two years have kept steady around approx. 59 mNOK pa. The major cost driver (past two years) under the section "equipment" has been lab rent, as Kavli implemented the lab rent model in late 2020. Otherwise the personnel costs represent the largest section of the overall expenditures, representing 72% of the total costs for the entire centre period. Multiple post. Doctoral fellows, PhDs and researchers has been financed by the funding over the last ten years (see attachment in section "special report" for further details).

English version The most challenging goal of neuroscience is to understand how subjective experience arises out of distributed electrical activity in the brain. Neuroscientists have uncovered striking correlations between behaviour and neural activity but the distributed and interleaved nature of neural representations has prevented insight into the underlying computational operations. The Centre of Excellence (CoE) scheme of the Research Council of Norway (RCN) provides a one-time opportunity to set up a research programme of the magnitude and continuance required to identify these mechanisms. I propose to develop the Centre for Neural Computation (CNC) to pioneer the extraction of computational algorithms from the mammalian cortex. We shall use our recent discovery of grid cells as an access ramp. Because the matrix-like firing of these cells is generated within the brain, independently of specific sensory features, grid cells provide unprecedented access to coding in high-end cortices. I have assembled an expert team that comes together at CNC to decipher the fundamental codes of the space circuit - the mechanisms by which signals from grid cells and other cells are generated, transformed, stored and retrieved by local and global network operations. The timing would be optimal because a wide spectrum of large-scale cell type-specific recording technologies will become available through NORBRAIN, an RCN-funded national infrastructure that less than 6 months after its opening is emerging as one of the world's largest assemblies of equipment for advanced neural circuit analysis. The research plan will be accompanied by the foundation of a national research school and the first PhD programme on neural circuits in Norway. The consequences of understanding the brain at the algorithmic level are diverse and far-reaching, spanning from the diagnosis and prevention of a wide range of neurological and psychiatric diseases to the implementation of brain algorithms in tomorrow's computers. Norwegian version En av de største utfordringene i moderne nevrovitenskap er å finne ut hvordan subjektiv erfaring oppstår ut fra elektrisk aktivitet i hjernen. Hjerneforskere har oppdaget slående korrelasjoner mellom atferd og nevral aktivitet men fordi nevrale representasjoner er distribuerte, har det vært vanskelig å oppnå innsikt i de underliggende beregningsoperasjonene. Forskningsrådets ordning for Fremragende Forskning representerer en unik anledning til å sette opp et forskningsprogram med den størrelse og varighet som trengs for å idenfisere slike mekanismer. Vi foreslår å opprette Centre for Neural Computation (CNC) for å lede an i avdekkingen av nevrale algoritmer i hjernebarken til pattedyr. Som innfallsport vil vi bruke gittercellene i entorhinal cortex, som vi oppdaget i 2005. Gitterceller er på mange måter målestokken i hjernens kartsystem. Fordi gittercellenes matriselignende aktivitetsmønster genereres inne i hjernen, uavhengig av innkommende sensoriske signaler, gir disse cellene oss en unik anledning til å beskrive prinsipper for nevral koding i høyere-ordens assosiasjons-hjernebark. Vi har samlet en gruppe eksperter for å identifisere de fundamentale kodene i hjernens kartsystem og for å avsløre mekanismene for dannelse og omdannelse av gitterceller og andre celler ved hjelp av lokale og globale nettverksoperasjoner. Tidspunktet for opprettelsen av et slikt senter er optimalt siden et vidt spektrum av storskala celletype-spesifikke registreringsteknologier har blitt tilgjengeliggjort gjennom NORBRAIN, en Forskningsrådsfinansiert nasjonal infrastruktur som 6 måneder etter innvielsen framstår som en av verdens største samlinger av utstyr for avansert nevral nettverksanalyse. Forskningsplanen ledsages av etableringen av en nasjonal forskerskole og det første PhD-programmet for nevrale kretser og nevral nettverksberegning i Norge. Implikasjonene av å forstå hjernen på et algoritmisk nivå er vidtrekkende og spenner fra tidlig diagnose og behandling av en rekke nevrologiske og psykiatriske sykdommer til utvikling av hjerne-computer-grensesnitt og implementering av nevrale algoritmer i fremtidens datamaskiner.