ERC FUNDED PROJECTS

"When I asked my seven-year-old daughter ""Who is the boy in your class who was also new in school last year, like you?"", she instantly replied ""Daniel"", using the descriptive content in my utterance to identify an entity in the real world and refer to it. The ability to use language to refer to reality is crucial for humans, and yet it is very difficult to model. AMORE breaks new ground in Computational Linguistics, Linguistics, and Artificial Intelligence by developing a model of linguistic reference to entities implemented as a computational system that can learn its own representations from data. This interdisciplinary project builds on two complementary semantic traditions: 1) Formal semantics, a symbolic approach that can delimit and track linguistic referents, but does not adequately match them with the descriptive content of linguistic expressions; 2) Distributional semantics, which can handle descriptive content but does not associate it to individuated referents. AMORE synthesizes the two approaches into a unified, scalable model of reference that operates with individuated referents and links them to referential expressions characterized by rich descriptive content. The model is a distributed (neural network) version of a formal semantic framework that is furthermore able to integrate perceptual (visual) and linguistic information about entities. We test it extensively in referential tasks that require matching noun phrases (“the Medicine student”, “the white cat”) with entity representations extracted from text and images. AMORE advances our scientific understanding of language and its computational modeling, and contributes to the far-reaching debate between symbolic and distributed approaches to cognition with an integrative proposal. I am in a privileged position to carry out this integration, since I have contributed top research in both distributional and formal semantics. "

1 499 805 €

Start date: 2017-02-01, End date: 2022-01-31

Learning to read is probably one of the most exciting discoveries in our life. Using a longitudinal approach, the research proposed examines how the human brain responds to two major challenges: (a) the instantiation a complex cognitive function for which there is no genetic blueprint (learning to read in a first language, L1), and (b) the accommodation to new statistical regularities when learning to read in a second language (L2). The aim of the present research project is to identify the neural substrates of the reading process and its constituent cognitive components, with specific attention to individual differences and reading disabilities; as well as to investigate the relationship between specific cognitive functions and the changes in neural activity that take place in the course of learning to read in L1 and in L2. The project will employ a longitudinal design. We will recruit children before they learn to read in L1 and in L2 and track reading development with both cognitive and neuroimaging measures over 24 months. The findings from this project will provide a deeper understanding of (a) how general neurocognitive factors and language specific factors underlie individual differences – and reading disabilities– in reading acquisition in L1 and in L2; (b) how the neuro-cognitive circuitry changes and brain mechanisms synchronize while instantiating reading in L1 and in L2; (c) what the limitations and the extent of brain plasticity are in young readers. An interdisciplinary and multi-methodological approach is one of the keys to success of the present project, along with strong theory-driven investigation. By combining both we will generate breakthroughs to advance our understanding of how literacy in L1 and in L2 is acquired and mastered. The research proposed will also lay the foundations for more applied investigations of best practice in teaching reading in first and subsequent languages, and devising intervention methods for reading disabilities.

2 487 000 €

Start date: 2012-05-01, End date: 2017-04-30

The potential and boundaries of the human mind is determined by dynamic interactions between the environment and the individual genetic architecture. However, despite several breakthroughs, the genetic revolution has not provided a coherent account of the development of the mind and its disorders, and the missing heritability is large across human traits. One explanation of this impasse is the complexity of the gene-environment interactions. Current knowledge about the determinants of a healthy mind is largely based on studies whose modus operandi is to treat the environment as a static entity, neglecting to consider the crucial fact that environmental inputs and their genetic interactions vary dramatically between life phases. The objective of BRAINMINT is to provide this missing link by zeroing in on two major life transitions, namely adolescence and pregnancy. These phases are characterized by temporarily increased brain plasticity, offering windows for adaptation and growth, but also host the emergence of common mental disorders. I propose that a multi-level investigation with this dark side of brain plasticity as the axis mundi will add a mechanistic understanding of this link between growth and vulnerability. I will test the main hypothesis that mechanisms that boost neuroplasticity promote adaptation to a dynamic environment, but at the cost of increased risk of psychopathology if exposed to a combination of genetic and environmental triggers. To this end I will utilize cutting-edge longitudinal brain imaging, electrophysiology, rich cognitive and clinical data, immune markers, gene expression and genetics. I will leverage on massive imaging data (n>40,000) and novel tools to increase power and generalizability and improve brain- and gene-based predictions of complex traits. Aiming to help resolving one of the modern day enigmas, BRAINMINT is a pioneering and high risk/high gain effort to find mechanisms of brain plasticity that support and harm the brain.

1 446 113 €

Start date: 2019-08-01, End date: 2024-07-31

Understanding how values of different options that lead to choice are represented in the brain is a basic scientific question with far reaching implications. I recently showed that by the mere-association of a cue and a button press we could influence preferences of snack food items up to two months following a single training session lasting less than an hour. This novel behavioural change manipulation cannot be explained by any of the current learning theories, as external reinforcement was not used in the process, nor was the context of the decision changed. Current choice theories focus on goal directed behaviours where the value of the outcome guides choice, versus habit-based behaviours where an action is repeated up to the point that the value of the outcome no longer guides choice. However, in this novel task training via the involvement of low-level visual, auditory and motor mechanisms influenced high-level choice behaviour. Thus, the far-reaching goal of this project is to study the mechanism, by which low-level sensory, perceptual and motor neural processes underlie value representation and change in the human brain even in the absence of external reinforcement. I will use behavioural, eye-gaze and functional MRI experiments to test how low-level features influence the neural representation of value. I will then test how they interact with the known striatal representation of reinforced behavioural change, which has been the main focus of research thus far. Finally, I will address the basic question of dynamic neural plasticity and if neural signatures during training predict long term success of sustained behavioural change. This research aims at a paradigmatic shift in the field of learning and decision-making, leading to the development of novel interventions with potential societal impact of helping those suffering from health-injuring behaviours such as addictions, eating or mood disorders, all in need of a long lasting behavioural change.

1 500 000 €

Start date: 2017-01-01, End date: 2021-12-31