Project acronym AgeConsolidate
Project The Missing Link of Episodic Memory Decline in Aging: The Role of Inefficient Systems Consolidation
Researcher (PI) Anders Martin FJELL
Host Institution (HI) UNIVERSITETET I OSLO
Call Details Consolidator Grant (CoG), SH4, ERC-2016-COG
Summary Which brain mechanisms are responsible for the faith of the memories we make with age, whether they wither or stay, and in what form? Episodic memory function does decline with age. While this decline can have multiple causes, research has focused almost entirely on encoding and retrieval processes, largely ignoring a third critical process– consolidation. The objective of AgeConsolidate is to provide this missing link, by combining novel experimental cognitive paradigms with neuroimaging in a longitudinal large-scale attempt to directly test how age-related changes in consolidation processes in the brain impact episodic memory decline. The ambitious aims of the present proposal are two-fold:
(1) Use recent advances in memory consolidation theory to achieve an elaborate model of episodic memory deficits in aging
(2) Use aging as a model to uncover how structural and functional brain changes affect episodic memory consolidation in general
The novelty of the project lies in the synthesis of recent methodological advances and theoretical models for episodic memory consolidation to explain age-related decline, by employing a unique combination of a range of different techniques and approaches. This is ground-breaking, in that it aims at taking our understanding of the brain processes underlying episodic memory decline in aging to a new level, while at the same time advancing our theoretical understanding of how episodic memories are consolidated in the human brain. To obtain this outcome, I will test the main hypothesis of the project: Brain processes of episodic memory consolidation are less effective in older adults, and this can account for a significant portion of the episodic memory decline in aging. This will be answered by six secondary hypotheses, with 1-3 experiments or tasks designated to address each hypothesis, focusing on functional and structural MRI, positron emission tomography data and sleep experiments to target consolidation from different angles.
Summary
Which brain mechanisms are responsible for the faith of the memories we make with age, whether they wither or stay, and in what form? Episodic memory function does decline with age. While this decline can have multiple causes, research has focused almost entirely on encoding and retrieval processes, largely ignoring a third critical process– consolidation. The objective of AgeConsolidate is to provide this missing link, by combining novel experimental cognitive paradigms with neuroimaging in a longitudinal large-scale attempt to directly test how age-related changes in consolidation processes in the brain impact episodic memory decline. The ambitious aims of the present proposal are two-fold:
(1) Use recent advances in memory consolidation theory to achieve an elaborate model of episodic memory deficits in aging
(2) Use aging as a model to uncover how structural and functional brain changes affect episodic memory consolidation in general
The novelty of the project lies in the synthesis of recent methodological advances and theoretical models for episodic memory consolidation to explain age-related decline, by employing a unique combination of a range of different techniques and approaches. This is ground-breaking, in that it aims at taking our understanding of the brain processes underlying episodic memory decline in aging to a new level, while at the same time advancing our theoretical understanding of how episodic memories are consolidated in the human brain. To obtain this outcome, I will test the main hypothesis of the project: Brain processes of episodic memory consolidation are less effective in older adults, and this can account for a significant portion of the episodic memory decline in aging. This will be answered by six secondary hypotheses, with 1-3 experiments or tasks designated to address each hypothesis, focusing on functional and structural MRI, positron emission tomography data and sleep experiments to target consolidation from different angles.
Max ERC Funding
1 999 482 €
Duration
Start date: 2017-05-01, End date: 2022-04-30
Project acronym AMORE
Project A distributional MOdel of Reference to Entities
Researcher (PI) Gemma BOLEDA TORRENT
Host Institution (HI) UNIVERSIDAD POMPEU FABRA
Call Details Starting Grant (StG), SH4, ERC-2016-STG
Summary "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.
"
Summary
"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.
"
Max ERC Funding
1 499 805 €
Duration
Start date: 2017-02-01, End date: 2022-01-31
Project acronym FASTPARSE
Project Fast Natural Language Parsing for Large-Scale NLP
Researcher (PI) Carlos GÓMEZ RODRÍGUEZ
Host Institution (HI) UNIVERSIDADE DA CORUNA
Call Details Starting Grant (StG), SH4, ERC-2016-STG
Summary The popularization of information technology and the Internet has resulted in an unprecedented growth in the scale at which individuals and institutions generate, communicate and access information. In this context, the effective leveraging of the vast amounts of available data to discover and address people's needs is a fundamental problem of modern societies.
Since most of this circulating information is in the form of written or spoken human language, natural language processing (NLP) technologies are a key asset for this crucial goal. NLP can be used to break language barriers (machine translation), find required information (search engines, question answering), monitor public opinion (opinion mining), or digest large amounts of unstructured text into more convenient forms (information extraction, summarization), among other applications.
These and other NLP technologies rely on accurate syntactic parsing to extract or analyze the meaning of sentences. Unfortunately, current state-of-the-art parsing algorithms have high computational costs, processing less than a hundred sentences per second on standard hardware. While this is acceptable for working on small sets of documents, it is clearly prohibitive for large-scale processing, and thus constitutes a major roadblock for the widespread application of NLP.
The goal of this project is to eliminate this bottleneck by developing fast parsers that are suitable for web-scale processing. To do so, FASTPARSE will improve the speed of parsers on several fronts: by avoiding redundant calculations through the reuse of intermediate results from previous sentences; by applying a cognitively-inspired model to compress and recode linguistic information; and by exploiting regularities in human language to find patterns that the parsers can take for granted, avoiding their explicit calculation. The joint application of these techniques will result in much faster parsers that can power all kinds of web-scale NLP applications.
Summary
The popularization of information technology and the Internet has resulted in an unprecedented growth in the scale at which individuals and institutions generate, communicate and access information. In this context, the effective leveraging of the vast amounts of available data to discover and address people's needs is a fundamental problem of modern societies.
Since most of this circulating information is in the form of written or spoken human language, natural language processing (NLP) technologies are a key asset for this crucial goal. NLP can be used to break language barriers (machine translation), find required information (search engines, question answering), monitor public opinion (opinion mining), or digest large amounts of unstructured text into more convenient forms (information extraction, summarization), among other applications.
These and other NLP technologies rely on accurate syntactic parsing to extract or analyze the meaning of sentences. Unfortunately, current state-of-the-art parsing algorithms have high computational costs, processing less than a hundred sentences per second on standard hardware. While this is acceptable for working on small sets of documents, it is clearly prohibitive for large-scale processing, and thus constitutes a major roadblock for the widespread application of NLP.
The goal of this project is to eliminate this bottleneck by developing fast parsers that are suitable for web-scale processing. To do so, FASTPARSE will improve the speed of parsers on several fronts: by avoiding redundant calculations through the reuse of intermediate results from previous sentences; by applying a cognitively-inspired model to compress and recode linguistic information; and by exploiting regularities in human language to find patterns that the parsers can take for granted, avoiding their explicit calculation. The joint application of these techniques will result in much faster parsers that can power all kinds of web-scale NLP applications.
Max ERC Funding
1 481 747 €
Duration
Start date: 2017-02-01, End date: 2022-01-31
Project acronym GEOCOG
Project Cognitive Geometry: Deciphering neural concept spaces and engineering knowledge to empower smart brains in a smart society
Researcher (PI) Christian Fritz Andreas DOELLER
Host Institution (HI) NORGES TEKNISK-NATURVITENSKAPELIGE UNIVERSITET NTNU
Call Details Consolidator Grant (CoG), SH4, ERC-2016-COG
Summary Through smart technology, we are overwhelmed with new information. Does this unlimited access to knowledge make us smarter? One of the key challenges for modern societies is to understand how the brain assembles our rich inventory of knowledge. Here, I will test the hypothesis that newly acquired knowledge is represented in the hippocampal formation in neural concept spaces, which are based on the coding principles and representational structures of the neural machinery involved in spatial navigation. The key idea is that the brain’s navigation system provides the building blocks of a neural metric for knowledge. In this groundbreaking cognitive neuroscience framework, I will bridge and integrate principles from Nobel Prize awarded neurophysiology and concepts from cognitive science and philosophy. Partly building on my ERC-StG project in which I discovered the core neural mechanisms underlying reconfiguration, integration and scaling of memory networks, the aim of my proposal is two-fold: 1. I seek to decipher neural concept spaces and unravel the neural codes of a cognitive geometry for knowledge and its deformations. 2. I will provide a proof-of-principle framework for next-generation neurocognitive technology and neural user models for cognitive enhancement to edit memories and engineer knowledge. Novel ‘Wikipedia’ learning tasks will be combined with state-of-the-art pattern analyses of space-resolved fMRI and time-resolved MEG to map and quantify representational structures. I will further develop AI-inspired analyses and closed loop brain-computer interfaces to perturb and edit neural concept space. The integrative mission of my program, from cells to systems-level involvement in cognition and to technology, opens up the exciting possibility to lay the ground for redefining cognitive neuroscience of knowledge by unravelling the fundamental neural principles of a cognitive topography and to make critical translations to empower smart brains in a smart society.
Summary
Through smart technology, we are overwhelmed with new information. Does this unlimited access to knowledge make us smarter? One of the key challenges for modern societies is to understand how the brain assembles our rich inventory of knowledge. Here, I will test the hypothesis that newly acquired knowledge is represented in the hippocampal formation in neural concept spaces, which are based on the coding principles and representational structures of the neural machinery involved in spatial navigation. The key idea is that the brain’s navigation system provides the building blocks of a neural metric for knowledge. In this groundbreaking cognitive neuroscience framework, I will bridge and integrate principles from Nobel Prize awarded neurophysiology and concepts from cognitive science and philosophy. Partly building on my ERC-StG project in which I discovered the core neural mechanisms underlying reconfiguration, integration and scaling of memory networks, the aim of my proposal is two-fold: 1. I seek to decipher neural concept spaces and unravel the neural codes of a cognitive geometry for knowledge and its deformations. 2. I will provide a proof-of-principle framework for next-generation neurocognitive technology and neural user models for cognitive enhancement to edit memories and engineer knowledge. Novel ‘Wikipedia’ learning tasks will be combined with state-of-the-art pattern analyses of space-resolved fMRI and time-resolved MEG to map and quantify representational structures. I will further develop AI-inspired analyses and closed loop brain-computer interfaces to perturb and edit neural concept space. The integrative mission of my program, from cells to systems-level involvement in cognition and to technology, opens up the exciting possibility to lay the ground for redefining cognitive neuroscience of knowledge by unravelling the fundamental neural principles of a cognitive topography and to make critical translations to empower smart brains in a smart society.
Max ERC Funding
2 000 000 €
Duration
Start date: 2017-05-01, End date: 2022-04-30
Project acronym InStance
Project Intentional stance for social attunement
Researcher (PI) Agnieszka Anna Wykowska
Host Institution (HI) FONDAZIONE ISTITUTO ITALIANO DI TECNOLOGIA
Call Details Starting Grant (StG), SH4, ERC-2016-STG
Summary In daily social interactions, we constantly attribute mental states, such as beliefs or intentions, to other humans – to understand and predict their behaviour. Today we also routinely interact with artificial agents: from Apple’s Siri to GPS navigation systems. In the near future, we will casually interact with robots. However, since we consider artificial agents to have no mental states, we tend to not attune socially with them in the sense of activating our mechanisms of social cognition. This is because it seems pointless to socially attune to something that does not carry social meaning (mental content) under the surface of an observed behaviour. INSTANCE will break new ground in social cognition research by identifying factors that influence attribution of mental states to others and social attunement with humans or artificial agents. The objectives of INSTANCE are to (1) determine parameters of others’ behaviour that make us attribute mental states to them, (2) explore parameters relevant for social attunement, (3) elucidate further factors – culture and experience – that influence attribution of mental states to agents and, thereby social attunement. INSTANCE’s objectives are highly relevant not only for fundamental research in social cognition, but also for the applied field of social robotics, where robots are expected to become humans’ social companions. Indeed, if we do not attune socially to artificial agents viewed as mindless machines, then robots may end up not working well enough in contexts where interaction is paramount. INSTANCE’s unique approach combining cognitive neuroscience methods with real-time human-robot interaction will address the challenge of social attunement between humans and artificial agents. Subtle features of robot behaviour (e.g., timing or pattern of eye movements) will be manipulated. The impact of such features on social attunement (e.g., joint attention) will be examined with behavioural, neural and physiological measures.
Summary
In daily social interactions, we constantly attribute mental states, such as beliefs or intentions, to other humans – to understand and predict their behaviour. Today we also routinely interact with artificial agents: from Apple’s Siri to GPS navigation systems. In the near future, we will casually interact with robots. However, since we consider artificial agents to have no mental states, we tend to not attune socially with them in the sense of activating our mechanisms of social cognition. This is because it seems pointless to socially attune to something that does not carry social meaning (mental content) under the surface of an observed behaviour. INSTANCE will break new ground in social cognition research by identifying factors that influence attribution of mental states to others and social attunement with humans or artificial agents. The objectives of INSTANCE are to (1) determine parameters of others’ behaviour that make us attribute mental states to them, (2) explore parameters relevant for social attunement, (3) elucidate further factors – culture and experience – that influence attribution of mental states to agents and, thereby social attunement. INSTANCE’s objectives are highly relevant not only for fundamental research in social cognition, but also for the applied field of social robotics, where robots are expected to become humans’ social companions. Indeed, if we do not attune socially to artificial agents viewed as mindless machines, then robots may end up not working well enough in contexts where interaction is paramount. INSTANCE’s unique approach combining cognitive neuroscience methods with real-time human-robot interaction will address the challenge of social attunement between humans and artificial agents. Subtle features of robot behaviour (e.g., timing or pattern of eye movements) will be manipulated. The impact of such features on social attunement (e.g., joint attention) will be examined with behavioural, neural and physiological measures.
Max ERC Funding
1 499 937 €
Duration
Start date: 2017-05-01, End date: 2022-04-30
Project acronym SIMULTAN
Project Aging-related changes in brain activation and deactivation during cognition: novel insights into the physiology of the human mind from simultaneous PET-fMRI imaging
Researcher (PI) Anna RIECKMANN
Host Institution (HI) UMEA UNIVERSITET
Call Details Starting Grant (StG), SH4, ERC-2016-STG
Summary There is no doubt that functional magnetic resonance imaging (fMRI) has led to a breakthrough in our ability to measure how the complexities of the mind are rooted in biology. However, deactivation of certain brain areas during cognitive control and increased activation of prefrontal areas in aging are two examples of consistently found patterns of fMRI activation that have had a large impact on the study of the human mind, but that prompt major questions of interpretation. The physiological basis of the fMRI signal reflects interplay between hemodynamics and metabolic demands that vary across the brain, as well as between different tasks and individuals, and cannot be resolved by fMRI alone. To be able to use non-invasive imaging to distinguish a normally aging brain from one that is in the pre-clinical stages of disease, it is important to understand the neurobiological basis of these functional brain changes. Positron emission tomography (PET) is a molecular imaging method that is able to monitor brain glucose metabolism, which stems primarily from synaptic activity and is invariant to changes in blood flow. Studies that have made use of the complementary information gained from fMRI and PET to investigate human brain function have had to rely on sequential scans, and correlation of the signals from both modalities between individuals. The investigation of within-person switches between different mental states with complementary modalities is only made possible by the recent development of a hybrid PET-MR system, which, for the first time, allows simultaneous assessment of fMRI signal, blood flow and PET glucose metabolism during cognitive task performance. The proposal is structured in three work packages that include PET-fMRI scans in 30 healthy younger and 40 older adults. The analyses are designed to disentangle the hemodynamic and metabolic contributions to fMRI deactivations and prefrontal over-activation in aging during cognitive task performance.
Summary
There is no doubt that functional magnetic resonance imaging (fMRI) has led to a breakthrough in our ability to measure how the complexities of the mind are rooted in biology. However, deactivation of certain brain areas during cognitive control and increased activation of prefrontal areas in aging are two examples of consistently found patterns of fMRI activation that have had a large impact on the study of the human mind, but that prompt major questions of interpretation. The physiological basis of the fMRI signal reflects interplay between hemodynamics and metabolic demands that vary across the brain, as well as between different tasks and individuals, and cannot be resolved by fMRI alone. To be able to use non-invasive imaging to distinguish a normally aging brain from one that is in the pre-clinical stages of disease, it is important to understand the neurobiological basis of these functional brain changes. Positron emission tomography (PET) is a molecular imaging method that is able to monitor brain glucose metabolism, which stems primarily from synaptic activity and is invariant to changes in blood flow. Studies that have made use of the complementary information gained from fMRI and PET to investigate human brain function have had to rely on sequential scans, and correlation of the signals from both modalities between individuals. The investigation of within-person switches between different mental states with complementary modalities is only made possible by the recent development of a hybrid PET-MR system, which, for the first time, allows simultaneous assessment of fMRI signal, blood flow and PET glucose metabolism during cognitive task performance. The proposal is structured in three work packages that include PET-fMRI scans in 30 healthy younger and 40 older adults. The analyses are designed to disentangle the hemodynamic and metabolic contributions to fMRI deactivations and prefrontal over-activation in aging during cognitive task performance.
Max ERC Funding
1 499 544 €
Duration
Start date: 2017-06-01, End date: 2022-05-31
