Cerca
 search for information or people
 
CIMeC - Center for Mind/Brain Sciences

Colloquium Series

The CIMeC Colloquium Series is an annual set of invited talks given by leading researchers in the mind/brain sciences, both from Italy and abroad, aimed principally at our PhD Students. Given the multi-disciplinary backgrounds of the CIMeC students and researchers, the colloquia are aimed at a general scientific level rather than at a more specialized audience. For school credit, all first, second and third year students attend the Colloquia. First and second year students prepare an essay based on one of the Colloquia of their choice summarizing the Colloquium, critically assessing the claims made and discussing the Colloquium in a broader context.

Find out who’s speaking at the next CIMeC Colloquium

Seminars 2019

Train (Trentino Autism Initiative) Seminar

The seminar is co-organized by the Center for Neuroscience and Cognitive Systems (CNCS) of IIT-UniTn and TRAIN (Trentino Autism Initiative)

Title: New therapeutic approaches to treat brain disorders characterized by impaired chloride homeostasis

When: May 28th, 2019 11 a.m.

Where: Conference room Palazzo Fedrigotti, 1st floor - corso Bettini 31, Rovereto

Speaker:  Dr. Laura Cancedda (IIT-Genova) 

 

Train (Trentino Autism Initiative) Seminar

When: Tuesday April 2nd, 2019, 5.00 P.M.
 
WhereSala Convegni Palazzo Fedrigotti, 1st floor (Rovereto)
 
Speaker: Prof. Martien Kas (University of Groningen, The Netherlands)
 
Title: Train seminar
Summary Prof. Martien Kas (University of Groningen, The Netherlands) is an internationally known behavioural neuroscientist, with a long experience in genetic and animal model studies of neurodevelopmental disorders, including autism. He will give a lecture entitled "From Autism Spectrum Disorders to quantitative biology; a neurodevelopmental perspective”. In his lecture, he will give a a comprehensive overview on quantitative and neurodevelopmental studies to address the complexity of autism pectrum Disorders.
 

CIMeC CLIC Seminar

When: Wednesday 6th February 2019 2.00 P.M. - 3.30 P.M

Where: Sala Convegni, First Floor, Palazzo Fedrigotti, Corso Bettini 31, Rovereto

Speaker: Sina Zarrieß, University of Bielefeld

Title: Modeling reference games on real-world images: A common playground  for computational and experimental methods

SummaryFor a long time, so-called reference games (cf. Rosenberg and Cohen, 1964) have been a popular experimental paradigm in linguistics, as they put together the basic ingredients of verbal interaction in a controlled microcosm: a speaker, a listener, a visual scene and a common communicative goal (identifying an object). Recently, this setting has been re-discovered in the Vision & Language community, and large-scale data sets that pair referring expressions with objects in real-world images have become available. These now constitute an excellent testbed for analyzing state-of-the-art machine learning models from a linguistic perspective and vice versa, and this is what I will do in this talk. First, I will present a computational study on neural referring expression generation, investigating different decoding strategies and how they relate to theoretical expectations. Second, I will present an ongoing eye-tracking study that leverages real-world images and a data-driven model for experimental hypothesis testing. Finally, I will discuss ideas on how to bring these two approaches even closer together.

Lab: CLIC

Host: Aurelie Herbelot

Seminars 2018

CIMeC Seminar

​When: Thursday, 22nd November 2018, 4.00 P.M. - 5.30 P.M

Where: Conference Room, First Floor, Palazzo Fedrigotti, Corso Bettini 31, Rovereto

Speaker: Afra Alishahi - Associate Professor, Tilburg Center for Cognition and Communication - https://ilk.uvt.nl/~aalishah/

​Title: Getting closer to reality: grounding and interaction in models of human language acquisition

​Summary: Humans learn to understand speech from weak and noisy supervision: they manage to extract structure and meaning from speech by simply being exposed to utterances situated and grounded in their daily sensory experience. Emulating this remarkable skill has been the goal of numerous studies; however researchers have often used severely simplified settings where either the language input or the extralinguistic sensory input, or both, are small scale and symbolically represented. I present a series of studies on modelling visually grounded language understanding. Using variations of recurrent neural networks to model the temporal nature of spoken language, we examine how form and meaning-based linguistic knowledge emerges from the input signal. Another aspect of learning a language which is often ignored in computational models of human language acquisition is the role of interaction. Children are not passive learners: they actively participate and sometimes guide conversation with their parents, and often influence the linguistic input produced by adults. I will present ongoing work which attempts to study the role of interaction in language learning, and to simulate a curiosity-based approach by the language learner to initiate topics of conversation and to improve the efficiency of the learning process.

Host: Raffaella Bernardi

CIMeC Seminar

​When: Tuesday, 6th November 2018, 10.30 A.M.

Where: Conference Room, First Floor, Palazzo Fedrigotti, Corso Bettini 31, Rovereto

Speaker: Leila Wehbe, Machine Learning Department, Carnegie Mellon University - http://www.cs.cmu.edu/~lwehbe/

​Title: What do naturalistic language experiments offer us?

​Summary: Progress in machine learning has allowed us to supplement hypothesis-driven science with data-driven science. In neuroscience, most language experiments to date have studied the brain by crafting conditions designed to isolate a single hypothesis. A more recent approach has focused instead on naturalistic language experiments in which subjects process a rich text. In this talk, I will describe progress in modeling language processing in the brain during naturalistic tasks, by using machine learning and natural language processing techniques to discover and test multiple hypotheses. I will also discuss recent projects that contrast results from naturalistic experiments with results from experiments with more constrained materials. I will compare results from running these different types of experiments on the same subjects. I will also showcase a reproducibility tool that we have developed to allow the community to test – via simulation – whether the results of published controlled experiments generalize to the naturalistic setting.

Host: Raffaella Bernardi

CIMeC Seminar

​When: Friday, 26th October 2018, 11.00 A.M.

Where: ACN lab - Seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speakers: Matthias Carl - Assistant Prof. of Physiology, Laboratory of Translational Neurogenetics, Center for integrative Biology, CIBIO

​Title: Why is every brain left-right asymmetric?

Host: Giorgio Vallortigara

CIMeC Seminar

​When: Wednesday, 10th October 2018, 11.00 A.M.

Where: ACN lab - Seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speakers: Hannah Rowland, Lorian Schweikert, Fabio Miazzi – Max Planck Institute for Chemical Ecology (Germany)

​Title: How toxic prey influence the evolution of predator behaviour

​Summary: Prey animals often advertise their chemical defences to predators by a distinctive and conspicuous visual appearance known as aposematism. Aposematic prey are typified by the red and black coloration of ladybirds, and the black and yellow stripes of cinnabar moth caterpillars, Tyria jacobaeae. My research investigates how the chemical and visual signals and chemical defences of aposematic prey affect the evolution of predator senses, psychology, and physiology. I will present research from laboratory and field experiments that investigate how birds perceive both taste and toxicity, how defence chemicals influence aversion learning, and how birds use social information about prey quality to make strategic-decisions about when to eat toxic prey.

Host: Uwe Mayer

CIMeC Seminar

When: Tuesday, 2nd October 2018, 4.00 P.M.

Where: ACN lab - Seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Giandomenico Iannetti - Professor of Neuroscience at University College London (UK) - https://www.iannettilab.net/

​Title: From pain to defensive actions: saliency detection as a reactive process

​Summary: The nervous system relates us to the rest of the world through perception and action: environmental information is continuously used to make decisions resulting in actions appropriate to achieve the ultimate objectives of life, survival and reproduction. For this reason, nervous systems are particularly sensitive towards the detection of salient environmental events that need to be rapidly acted upon and imperil survival - a typical example being transient nociceptive stimuli causing pain. These stimuli elicit extremely large brain responses, which have been traditionally used to build models of where and how painful percepts are generated in the human brain, and, more recently, to infer whether an individual is in pain. I will provide evidence that this dominant view is incorrect. Instead, I will suggest that the largest part of these brain responses reflect a basic mechanism through which the human brain detects and purposefully reacts to behaviourally-relevant sensory events, regardless of their perceptual quality. I will describe a basic physiological mechanisms that couples these saliency-related cortical responses with an activation of the motor system, indicating that saliency detection is not merely perceptive but reactive, preparing the animal for subsequent appropriate actions. I will finally show how stimuli occurring near the body elicit stronger behavioural and physiological responses. This phenomenon, which makes evolutionary sense (a predator within striking distance is more salient than one farther away), led to the concept of peripersonal space (PPS). The common and intuitive description of PPS as a single, distance-based, in-or-out zone, is however contradicted by empirical data. I propose a reconceptualization that incorporates PPS into mainstream theories of action selection and behaviour.

Host: Giorgio Vallortigara

CIMeC Seminar

​When: Tuesday, 18th September 2018, 10.30 - 11.30 A.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy, 1st floor conference room

Speaker: Dr. Irina Simanova

​Title: Words shape object representations in the brain

​Summary: Humans possess the unique ability to name objects. How does this ability transform cognition and perception? This question goes to the core of what it means to be human. Among philosophers and cognitive scientists, many have commented on the way verbal labels enable us to access and manipulate mental representations efficiently in a unique way. In this talk I will present my work on the interplay between verbal labels, concepts, and percepts. An important empirical question is: what kind of representations are activated by linguistic labels? In a recent study we addressed this question by studying how labels affect the processing of upcoming visual information. Specifically, we tested the hypothesis that words affect visual perception be pre-activating visual object shape.

Host: Scott Fairhall

CIMeC Seminar

​When: Monday, 17th September 2018, 3.00 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy, 1st floor conference room

Speaker: Prof. John K. Tsotsos (York University, Canada) 

​Title: Sampling the Attentional Spectrum within the Selective Tuning Model

​Summary: The Selective Tuning model of visual attention and visual processes has been under active development since 1987. During this period, the breadth of its functionality has been theoretically and experimentally investigated. This presentation will begin with a brief overview of its basic concepts and structure, from both computational as well as biological points of view.  Following this, a number of recent developments will be presented that show how Selective Tuning deals with many of the facets of attention in visual  tasks. These include attentional suppressive surround effects in novel stimulus types, how attention changes during the  life cycle in humans, fixation prediction that matches human performance, attentional networks for executive control, and spatial cognition in 3D for an active observer. 

Host: Luca Ronconi and David Melcher

CIMeC Seminar

​When: Friday, 7th September 2018, 10.30 - 12.00 A.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy, Conference Room

Speaker: Jonathan Cant, Psychology Department, University of Toronto Scarborough, Toronto, Canada

​Title: Making a scene: How visual features contribute to scene representation

​Summary: It has previously been demonstrated that the scene-sensitive parahippocampal place area (PPA) is more active for judgments of the surface texture and material properties of single isolated objects, compared to judgments of object shape. On the surface, this appears inconsistent with the view that PPA is specialized for processing scenes, since the single objects were not presented in the context of a scene. However, surface texture (and the material-properties signaled by texture) is important in scene processing as it can be used to aid in image segmentation, can contribute to the recognition of scene identity, and can provide affordance-related cues relevant for navigation. Thus, the finding that attending to object texture and material activates PPA may be better interpreted as evidence that PPA utilizes multiple visual features, in addition to its well-known role in processing global spatial features such as structural geometry, when representing scenes. Building on this observation, in this talk I will present the results of several studies examining the contribution of different visual features to scene representation in human occipito-temporal cortex. Specifically, I will present results investigating: 1) whether scene-selective cortex is more sensitive to processing scene, compared with object, texture; 2) whether the processing of different scene features (i.e., scene geometry/layout and scene texture/material) is mediated by shared or distinct neuronal mechanisms in scene-selective cortex; 3) whether the importance of scene layout and scene texture varies according the type of scene category being perceived (i.e., open vs. closed scenes, and natural vs. manufactured scenes); 4) how task context influences the representation of scenes in occipito-temporal cortex; and 5) the relation between object-ensemble perception and texture perception in scene processing. Finally, given that we do not typically perceive scenes devoid of objects (and in turn, we do not perceive objects outside of the context of a scene), I will present some recent results investigating interactions between scene and object perception (i.e., does global/scene perception interfere with local/object perception, or vice versa?). Taken together, these results will demonstrate that multiple visual features are represented in human scene-selective cortex, and that this representation is flexible, as the importance of different scene features varies according to perceived scene category and the goals of the observer. Moreover, the finding that object-scene interactions are influenced by both global and local image features may explain how one is able to perceive both the “entire forest” and the “individual trees” from a visual scene.

Host: Simona Monaco

CIMeC Talk

​When: Monday, 20th August 2018, 2.00 P.M.

Where: CIMeC via delle regole 101, Mattarello Conference Room

Speaker: Davide Marchiori, Associate professor (University of Southern Denmark)

​Title: Two destinies of hierarchical learning

​Summary: Human organizations are commonly characterized by a hierarchical chain of command that facilitates division of labor and integration of effort. Higher-level employees set the strategic frame that constrains lower-level employees who carry out the detailed operations serving to implement the strategy. Typically, strategy and operational decisions are carried out by different individuals that act over different timescales and rely on different kinds of information. We hypothesize that when such decision processes are hierarchically distributed among different individuals, they produce highly heterogeneous and strongly path-dependent joint learning dynamics. To test this, we design laboratory experiments of human dyads facing repeated joint tasks, in which one individual is assigned the role of carrying out strategy decisions and the other operational ones. The experimental behavior generates a puzzling bimodal performance distribution–some pairs learn, some fail to learn after a few periods. We also develop a computational model that closely mirrors the experimental settings and accurately predicts the heterogeneity of performance by human dyads. Comparison of experimental and simulation data suggests that self-reinforcing dynamics arising from initial random conditions are sufficient to explain the performance heterogeneity observed experimentally, without any need to resort to heterogeneous individual traits of the agents.

Host: Luca Polonio

CIMeC Talk

​When: Thursday, 26th July 2018, 10 - 11 A.M.

Where: CIMeC ACN lab, seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Prof. Mikolaj Hernik - Cognitive Development Center, Central European University

​Title: Communicative competence of human infants

​Summary: This talk will focus on some recent and ongoing studies investigating the hypothesis that human infants may be from very early on well prepared to be the recipients of ostensive referential communications. I will present our attempts to study questions concerning cultural universality of infants' early communicative competencies, and concerning the presence of genericity-bias in infants' encoding of the content of communicative demonstrations. I will also discuss current state of the debate concerning the role of ostension in infant gaze-following.

CIMeC Open Science Seminar

​When: Thursday, 19th July 2018, 4 - 5 P.M.

Where: CIMeC via delle regole 101, Mattarello Conference Room

Speaker: Prof. Davide Crepaldi, SISSA, Trieste, Italy

​Title: Implementing an open-everything policy in the lab

​Summary: Promoting a better science isn’t about fixing a few things in our usual research routine. In order to effectively share data and analysis codes, keep ourselves clear of bad practices and, more generally, carry out research more rigorously and transparently, we need to set up an appropriate pipeline from day 1, which guides us towards meeting all the requirements for a healthy science, gently and naturally. Even more importantly, we have to convince everyone in our labs to follow this pipeline - which can be surprisingly difficult, even with the best and more open-science prone students and postdocs. In this talk, I’ll share my personal experience in this respect - how this pipeline looks like in our lab, what is working smoothly and what is not, what are the advantages of having such a pipeline, and what are the biggest hurdles that we're facing in sticking to it.

Transfer-Learing Seminar

​When: Tuesday, 17th July 2018, 3 - 4 P.M.

Where: CIMeC via delle regole 101, Mattarello Conference Room

Speaker: Prof. Vittorio Pelligra, Associate Professor of Economics, University of Cagliari, Italy

​Title: Empathy, altruism and strategic behavior in simple economics games

​Summary: The ability to empathize is a fundamental social competence. Being able to ascribe intentions, desires, beliefs to other people (cognitive empathy), to understand and, to some extent, share their emotional states (affective empathy) is what, ultimately, renders us ‘social animals’, ‘relational beings’: Individuals capable of properly living in a community - able to anticipate others’ behavior in order to support their own interests but also to coordinate and cooperate with others to achieve mutual benefits. Despite its important role for understanding interpersonal behavior, economists and game theorists have devoted little attention to how empathy interacts with preferences, beliefs and choices. In this paper, we present results from a laboratory test devised to systematically investigate whether and how heterogeneity in individuals’ empathic abilities affects their behavior in a set of simple economic games. We find a positive impact of empathy on individuals’ generosity, sensitivity to inequality, trust and, finally, a significant inverse relationship between empathy and strategic sophistication. As a whole, our results lead to the conclusion that empathy tends to promote pro-social behaviors with no regard for strategic considerations.

Transfer Talk

​When: Tuesday, 10th July 2018, 3 P.M.

Where: CIMeC via delle regole 101, Mattarello Conference Room

Speaker: Eve Fabre, Institut Supérieur de l’Aéronautique et de l’Espace (ISAE) of Toulouse, France

​Title: Investigating the impact of information asymmetry between leaders and subordinates on group problem solving in a simulated space mission

​Summary: With the prospects of possible long-term missions to Mars, the study of group interaction and more specifically how to organize astronauts’ groups appears as essential to both the selection and the training of future crews. Each year, 7 space engineering students of the ISAE-supaéro are selected to take part in a simulated 3-week manned space mission in the Mars Desert Research Station (MDRS) in the Utah desert. This annual mission provides a unique opportunity to conduct social psychology experiments in a space-like environment. Participants played the computer game Keep Talking and Nobody Explodes © in groups of three. In this game, one player “the defuser” is placed in front of the computer and has to defuse a bomb composed of various modules. S/he receives the help of two spatially distant “readers” who own a manual containing the information necessary to defuse the bomb. Each triad was composed of two officers and one leader (i.e., the commander or the vice-commander of the mission) and played randomly as defuser or reader. The power of the readers (i.e., access to the information to defuse the bomb) was manipulated. In the control condition both manuals were correct, while in the condition of interest one of the manuals was correct and the other one was partly blacked-out to create a power asymmetry between the two readers. We investigated the impact of this information asymmetry on the behavior of both low-status (i.e., officers) and high-status participants (i.e., commander and vice-commander). The preliminary results on both speech and emotion analysis will be presented during the talk.

Host: Giorgio Coricelli

Seminar

​When: Monday, 9th July 2018, 5 - 6 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 3rd floor seminar room

Speaker: Marco Battaglini, Department of Neurological and Behavioral Sciences, University of Siena, Italy

​Title: Decreased accuracy of brain segmentation tools applied to the MRI of multiple sclerosis subjects: issues and possible solutions

​Summary: The seminar has the aim to illustrate the major issues arising with thesegmentation MRI of multiple sclerosis (MS) subjects: i) The misclassification of gray and white matter due to the deep differences between the global intensities histograms of MS patients and healthy controls; ii) the failure of available segmentation tools to reach the required accuracy needed to detect changes in atrophy MS patients and healthy controls. To face the first issue, the "filling lesions" solution will be presented and its effectiveness will be evaluated. To face the second issue, given a pair of MRI longitudinally acquired from the same subject, two distinct approaches will be presented: the first one based on the integration of Jacobian of non-linear registration coefficients between the pair of scans and the second one based on the equalization of the pair of intensity histograms of the two images. At the end of the seminar, it will be clear that: i) intensity changes in brain MRI of diseased subjects lead to changes in performances of MRI segmentation tools; and that ii) these decreased performances need to be taken in account, together with a-priori knowledge of the expected magnitude of the studied biomarker, to better design MRI studies.

Host: Jorge Jovicich, Ph.D.

Seminar

​When: Thursday, 28th June 2018, 3 - 4 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy conference room 

Speaker: Ilana Ritov, Professor of Psychology, Center for the Study of Rationality, The Hebrew University of Jerusalem

​Title: Effects of competition and social status

​Summary: Competition motivates individuals to work harder and obtain better outcomes. On the downside, it also increases the propensity of contestants to behave in a-social And even unethical manner before and during competition. However, the effect of engaging in a competition does not end when the competition is over. It spills over to later, unrelated decisions. Social status, often gained through competition has similar effects. I will present research in which we experimentally manipulate competition outcomes and examine post-competition behavior. In particular, we ask how competing and social status affect subsequent pro- and anti-social behavior, and who among winners and losers makes more selfish choices and engages in unethical behaviors.

Host: Katya Tentori

Seminar

​When: Wednesday, 27th June 2018, 4 - 5 P.M.

Where: CIMeC via delle regole 101, Mattarello, Conference room

Speaker: Gabriel Enrique Varela Mattatall, PhD Candidate, Biomedical Imaging Center at Pontificia, Universidad Catolica de Chile

​Title: Diffusion MRI from a signal processing perspective: undersampling and applications of q-space imaging.

CIMeC Seminar

​When: Tuesday, 26th June 2018, 3 -4 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 1st floor conference room

Speaker: Evan Livesey, University of Sydney

​Title: Preparing to act: It’s not what you expect

​Summary: Performing voluntary actions in everyday life requires a combination of proactive planning and reacting to changing circumstances, and is widely thought to be the remit of executive functions directed by our conscious expectations about future events. However, theorists have proposed that action selection and inhibition are often controlled by stimuli associated with past behaviour rather than our expectation of goal-relevant events in the future. Testing these potential influences on motor planning is important for theories of both associative learning and cognitive control. Recently, we have been using an analysis developed in Pavlovian conditioning, which demonstrates a strong dissociation between conditioned responding in anticipation of an outcome and the individual's explicit expectancy of that outcome. We have tested for this dissociation, referred to as the Perruchet effect, in a range of speeded response tasks with different demands on attention, response selection, and response inhibition. I will discuss some of the evidence we have found for this dissociation in rapid response tasks as well recent tests of its basis in associative learning using transcranial magnetic stimulation to probe motor preparation in primary motor cortex.

Host: Carlo Miniussi

CIMeC Talk

​When: Monday, 25th June 2018, 5 - 6.30 P.M.

Where: Manifattura CIMEC - ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14 

Speaker: Anna Wilkinson - University of Lincoln UK, School of Life Sciences, Animal Cognition

​Title: Factors impacting social cognition in reptiles: from robots to incubation 

​Summary: The transfer of complex behavioural information (as opposed to specific stimuli, such as isolated vocalisations) between individuals has been observed in a surprisingly wide variety of species. With strong evidence that many animals can, through observation alone, acquire sophisticated behavioural information from conspecifics accurately enough for them to replicate the behaviour themselves. However, variation in performance is frequently observed both within and between experiments. Generally, ‘noise’ in the information (caused, for example, by variation in the proficiency with which a demonstrator performs a task or variation in the information accessible to the naive animal, perhaps because of visual occlusion) is a factor largely outside the experimenter’s control, and often extremely difficult to quantify. Therefore, although animals can undoubtedly extract relevant information in order to learn demonstrated behaviours, it is often unclear (a) whether variation in learning ability stems from the demonstrator or the learner, and (b) exactly which specific aspects of a complex behavioural repertoire the learner is responding to. This talk will discuss two ways in which we have investigated this. The use of realistic robots can provide demonstrators to perform behaviours with sufficient accuracy and repeatability to allow others to learn from them, and so that all observers have the potential to receive the same information. Here, we describe our work on the construction of robotic bearded dragons (Pogona vitticeps), and their implementation in social behaviour and cognition experiments. Secondly we investigate the impact of early individual experience and the impact that this may have on learning later in life.

Host: Giorgio Vallortigara

Visiting Professor Lecture

​When: Monday, 25th June 2018, 3 -4 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 3rd floor seminar room

Speaker: Ilaria Berteletti, Assistant Professor

​Title: Educational Neuroscience: What is it? What it's not.

​Summary: In this lecture I will review the aim and scope of the emerging field known as Educational Neuroscience. In the past decade, several criticisms have been raised claiming that between Neuroscience and Education the "bridge is too far". I will go over some of the most common misconceptions raised by skeptics but also present a framework that will guide meaningful translation between the two fields.

Seminar

​When: Monday, 25th June 2018, 2 -3 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 1st floor conference room 

Speaker: Dr. Paul Bello, Director, Interactive Systems, Naval Research Laboratory USA

​Title: The attentional basis of quantity perception: theory and computation

Host: David Melcher

CIMeC Talk

​When: Thursday, 21th June 2018, 3 - 4 P.M.

Where: ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Prof.sa Anna Wilkinson - University of Lincoln UK, School of Life Sciences, Animal Cognition

​Title: Cold-Blooded Cognition: Learning and memory in reptiles

​Summary: Very little is known about the cognitive abilities of reptiles. They have traditionally been considered to be “sluggish and unintelligent creatures” (Yerkes 1901, p 520) and have largely been ignored in the study of animal cognition. However, more recent research has revealed an impressive suite of cognitive abilities. To gain an understanding of the evolution of cognition in amniotes, it is necessary to carry out direct experimental investigations of the learning and memory abilities of reptiles that parallel the extensive work already available in mammals and birds. Examination of the cognitive mechanisms underlying the behaviour of these animals can provide crucial information about the evolution of the brain. This talk will present some recent research on the cognitive abilities of reptiles and compare them to what is known about these processes in other animals.

Host: Giorgio Vallortigara

Seminar

​When: Friday, 15th June 2018, 11 - 12 A.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 2nd floor meeting room

Speaker: Ilaria Sani, Rockefeller University Leon Levy Post-Doctoral Fellow

​Title: Pathways of Attention

​Summary: To safely navigate their environment, people and animals must rapidly select, from the flood of incoming sensations, the pieces of information most relevant to their goals. But, when faced with new circumstances, they must quickly redirect their mental resources and change their course of action. Attention is the brain function that takes on this continuous challenge of selecting relevant information and directing resources, by linking the outside world with internal cognitive state. How the brain functionally and structurally implements this fine attentional control remains a deep mystery in cognitive neuroscience. Classical studies of visual attention, from neuropsychology to single unit physiology, have identified areas of parietal and frontal cortex like the lateral intraparietal area (LIP) and the Frontal Eye Field (FEF) as sources of attentional control. Recently, a ventral area in the macaque temporal cortex, the posterior infero-temporal dorsal area PITd, was found to exhibit strong attentional modulation and a functional profile similar to LIP and FEF. PITd was thus proposed to be a putative attentional control area. This raises the question of whether and how these spatially distant areas coordinate a joint focus of attention. Here we test the hypothesis that FEF, LIP, and PITd are directly interconnected. To determine connectivity between functionally specific attention areas, we combined functional MRI of attention areas with high-resolution diffusion MRI and probabilistic tractography. We found that LIP, FEF, and PITd are all directly connected to each other: LIP and FEF connect through the third branch of the Superior Longitudinal Fasciculus, PITd communicates with LIP through the vertical limb of the Inferior Longitudinal Fasciculus, with the FEF via the Extreme Capsule. These results ascribe a new function, the communication of endogenous attention signals, to two known fiber-bundles, they highlight the importance of vertical interactions across the two streams of the primate visual system, and imply that the control of endogenous attention, hitherto thought to reside in dorsal cortical areas, is exerted by a dorso-ventral cortical network, thus providing a new view onto the neural mechanisms of attentional control.

Host: David Melcher

CIMeC Seminar

​When: Thursday, 7th June 2018, 4 - 5 P.M.

Where: Mattarello - conference room, basement, Via delle Regole 101.

Speaker: prof. Michele Piana, coordinator of the MIDA Group - Methods for Image and Data Analysis, Dipartimento di Matematica, Università di Genova e CNR - SPIN, Genova

​Title: Computational methods for the modeling of neurophysiological data

MADVIS Seminar

​When: Monday, 4th June 2018, 3 - 4 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy 1st floor conference room 

Speaker: Elia Formisano, Maastricht Brain Imaging Center, Department of Cognitive Neuroscience, Maastricht University

​Title: From ears to brain (and back): Imaging the brain computations for sound analysis

​Summary: A friend speaking, a bird chirping, a piano playing. Any living being or non-living object that vibrates generates acoustic waveforms, which we call sounds. How does our brain transform these acoustic vibrations into meaningful percepts? This lecture illustrates current research combining computational modeling with functional magnetic resonance imaging (fMRI) at ultra-high magnetic field (7 and 9.4 Tesla) aimed at discovering the neuro-computational mechanisms for sound analysis in human auditory cortex.

Contact: Olivier Collignon

Clicker's Talk

​When: Wednesday, 30th May 2018, 3 - 4 P.M.

Where: 3rd floor seminar room CIMeC Fedrigotti

Speaker: Aurelie Herbelot

​Title: Speakers in vats: simulating model-theoretic alignment with distributional semantics

​Summary: One long-standing puzzle in semantics is the ability of speakers to refer successfully in spite of holding different models of the world. This puzzle is famously illustrated by the cup/mug example: if two speakers disagree on whether a specific entity is a cup or a mug (i.e. if their interpretation functions differ), how can they align so that the entity can still be talked about? Another puzzle, coming to us through lexical and distributional semantics, is that word meaning seems to be infinitely flexible, indeed much more so than the traditional notion of sense would have it. This makes the alignment process between speakers even more unpredictable. In this talk, I will report on a series of experiments aiming at investigating what differences in language use can tell us about the ability of speakers to align at a model-theoretic level. Since speaker-dependent data is extremely hard to obtain, I propose a new methodology to 'spawn' speakers from a reference distributional semantic space, corresponding to different types of variations in language use. I show how and where alignment is disturbed, and give a theoretical account of how such perturbations relate to potentially catastrophic differences in world representations.

CIMeC Visiting Professor Seminar

​When: Tuesday, 29th May 2018, 1.30 - 2.30 P.M.

Where: Palazzo Fedrigotti, Corso Bettini, 31, 38068 Rovereto TN, Italy

Speaker: Justin Harris, Professor, University of Sydney, Australia.

​Title: Stop the press! Response inhibition and motor cortex excitability

​Summary: An important component of response control is the ability to abort an action after it has been initiated, such as withholding an expletive when powerpoint corrupts your slides during a scientific talk. A simple but reliable test, the Stop Signal Task, shows that people vary considerably in how quickly or efficiently they can stop a response that has been initiated. I will describe a series of experiments from our lab that use Transcranial Magnetic Stimulation (TMS) to reveal how inhibitory processes in the motor cortex contribute to response stopping.

Host: Carlo Miniussi

Clic seminar

​When: Thursday, 24th May 2018, 3.00 - 4.30 P.M.

Where: CIMeC, 3rd floor seminar room Palazzo Fedrigotti 

Speaker: Grzegorgz Chrupala, Tilburg University

​Title: Neural representations of form and meaning in spoken language

​Summary: The task of learning language in a multisensory setting, with weak and noisy supervision, is of interest to scientists trying to understand the human mind as well as to engineers trying to build smart conversational agents or robots. In this talk I will present work on learning language from visually grounded speech using deep recurrent neural networks, and show that these models are able to extract linguistic knowledge at different levels of abstraction from the input signal. I then describe analytical methods which allow us to better understand the nature and localization or representations emerging in such recurrent neural networks. I will also discuss the challenges inherent in fully unsupervised modeling of spoken language and present recent results on this problem.

Contact: Aurelie Herbelot

TRANSFER talk

​When: Friday, 18th May 2018, 10.30 - 11.30 A.M.

Where: 3rd floor seminar room Corso Bettini 31 Rovereto Palazzo Fedrigotti

Speaker: James Tremewan, Research Fellow, Department of Economics, Vienna University.

​Title: Endogenous conflict of interest: the effects of transparency and communication

​Summary: There are two distinct perspectives to look at the relationship between human language and the brain, i.e. where and what neurons fire. The role of linguistic theory is essential in both cases: I will show that the identification of “Impossible languages”, that is of non-recursive syntax, and the recording of “the sound of thought” in inner speech provides two major examples of these two perspectives.

Contact: Giorgio Coricelli

ACN Talk

​When: Wednesday, 16th May 2018, 4 - 5 P.M.

Where: ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Andrea Moro, Full Professor of General Linguistics, Deputy Rector, University School for Advanced Study IUSS Pavia

​Title: Linguistics meets the Brain

​Summary: There are two distinct perspectives to look at the relationship between human language and the brain, i.e. where and what neurons fire. The role of linguistic theory is essential in both cases: I will show that the identification of “Impossible languages”, that is of non-recursive syntax, and the recording of “the sound of thought” in inner speech provides two major examples of these two perspectives.

Contact: Giorgio Vallortigara

Clicker's Talk

​When: Wednesday, 9th May 2018, 3 - 4 P.M.

Where: 3rd floor seminar room CIMeC Fedrigotti Corso bettini 31 Rovereto

Speaker: Raffaella Bernardi

​Title: See both the forest and the trees: Grounding by zooming out or zooming in

​Summary: Humans can see the forest, but can also see the trees, if they need to do so. Current language and vision models can be trained to either see the forest (but they would miss the trees) or to see its trees (but they would miss the forest). In other words they fail to zoom in and out. We show that their coarse representations are meaningful when fuzzy operations need to be performed, like set comparison or vague quantification, but are not suitable when evaluated on tasks like grounded textual entailment, that require a fine-grained representation. We conjecture that such models could gain a precise representation through grounded interaction. Through a dialogue about objects in an image, a model can learn to zoom into forest's trees and zoom out of them to see the forest, if the focus on the image has to be changed. To this end, we discuss our grounded conversational agents inspired by cognitive studies and the tradition of dialogue systems.

CLIC Seminar

​When: Wednesday, 2nd May 2018, 3.00 - 4.30 P.M.

Where: Seminar Room third floor (C302), CIMeC, Palazzo Fedrigotti, Corso Bettini, 31, Rovereto

Speaker: Denis Paperno (Loria, CNRS)

​Title: Predictability in the semantic space: towards a distributional characterization of the infection-derivation distinction

​Summary: The literature is divided on the relationship between inflection and derivation. Many morphologists, of various theoretical inclinations, hold that inflection and derivation are essentially the same thing (Robins, 1959; Di Sciullo & Williams, 1987; Bochner, 1993; Booij, 1996; Kœnig, 1999), or that there is at most a gradient distinction between canonical inflection and canonical derivation (Dressler, 1989; Corbett, 2010; Spencer, 2013). Many others hold on the contrary hold that there is an irreducible difference between the two (Matthews 1965, 1974; Anderson 1982; Perlmutter 1988; Aronoff 1994; Stump 2001). We focus on one of the criteria that are discussed in the literature, which we term stability of contrast. In a nutshell, inflection is supposed to be stable in its syntactic and semantic effects across lexemes (books is to book as cats is to cat), while derivation is expected to be less so (delegation is not to delegate as election is to elect). The idea that inflection and derivation differ in this way is intuitively compelling, and has been stated repeatedly (Robins 1959: 125-126; Matthews 1974, 49-52; Wurzel 1989, 36; Stump 1998). However, to our knowledge, no previous study has attempted to define stability of contrast in an operational fashion, and to test on a large scale the validity of a difference between inflection and derivation: rather, all studies discuss intuitive semantic contrasts between hand-picked series of pairs of words. To test the contrast stability of different morphological relations, we construct word form triples where R1 is an inflectional relation and R2 is derivational. By hypothesis, we expect the vector offsets for derivationally-related pairs to be more diverse than those for inflectionally-related pairs. We employ the Euclidian distance between the vector offsets and the mean vector for the same relation as our main measure of diversity. The null hypothesis in t-test analysis is that the means of the two paired samples are identical. The contrast stability hypothesis will be confirmed if the deviation from average is greater for derivational relations than for inflectional ones with a significant t-test value. Our experiment compares different kinds of relations between word forms in a controlled, paired-sample setting. The results support that there is a systematic contrast between derivational and inflectional relations. This finding fully agrees with existing theoretical literature and is meant to inform the ongoing debate on the status of derivation and inflection in language and cognition. (Based on joint work with Olivier Bonami and Timothee Mickus.)

Contact: Aurelie Herbelot

Seminar

​When: Thursday, 19th April 2018, 3.30 P.M. 

Where: Levico Room, DISI, Via Sommarive, 9, 38123 Povo, Trento

Speaker: Jasper Uijlings, research scientist at Google

​Title: Human-Machine collaboration for Large Scale Bounding Box Annotation

​Summary: Modern vision systems requires lots of manually annotated data for training, which is expensive to obtain. In this talk I will present our efforts at Google for efficiently annotating bounding boxes in images at scale using Human-Machine collaboration. First I will talk about our work on Knowledge Transfer (CVPR 18), where we train object detectors on target classes from weakly supervised training images, helped by a set of source classes with bounding-box annotations. We use this and several other techniques to improve upon our previous work (CVPR 16) in which humans are asked to verify an automatically generated bounding box, circumventing the need for manual drawing. Then I will describe an improved method for manually drawing boxes, called Extreme Clicking (ICCV 17). Finally, I describe how we train an agent to automatically select between box verification questions and manual drawing, resulting in Intelligent Annotation Dialogs (CVPR 18).

Contact: Raffaella Bernardi

CIMeC Open Science Seminar

​When: Tuesday, 27th March 2018, 10.30 A.M. - 12.00 P.M.

Where: CIMeC Palazzo Fedrigotti, 1st floor Seminar room, Corso Bettini, 31 Rovereto 

Speaker: Prof. Roberto Caso and Prof. Paolo Guarda from the Faculty of Law, University of Trento, Italy

​Title: Open Access Publishing Seminar

​Summary: The talk will cover different kinds of open access as well as academic copyright and creative commons licenses. Open access increases the pace of information flow as well as increasing our visibility amongst scientists and providing better science access to the general public. It is a core strategy of the European Commission as well as a key component of CIMeC's Strategic Plan.

CLIC Seminar

​When: Wednesday, 14th March 2018, 10.30 A.M. - 12.00 P.M.

Where: Ex-chiesetta (a.k.a. room C13, ground floor), CIMeC, Palazzo Fedrigotti, Corso Bettini, 31, Rovereto

Speaker: Marco Baroni (Facebook AI Research) / https://research.fb.com/people/baroni-marco/

​Title: Systematic compositionality with recurrent neural networks

​Summary: Recurrent neural networks (RNNs) are remarkably general learning systems that, given appropriate training examples, can handle complex sequential processing tasks, such as those frequently encountered in language and reasoning. However, RNNs are remarkably sample-heavy, typically requiring hundreds of thousands of examples to master tasks that humans can solve after seeing just a few exposures. The first set of experiments I will present shows that modern RNNs, just like their ancestors from the nineties, have problems with systematic compositionality, that is, the ability to extract general rules from the training data, and apply them to new examples. As systematic compositionality allows very fast generalization to unseen cases, lack of compositional learning might be one of the roots of RNN's training data thirst. I will next present an ongoing study where RNNs must solve an apparently simple task where correct generalization relies on function composition. Current results suggest that a large random search in RNN space finds a small portion of models that converged on a (limited) compositional solution. However, it's not clear, for the time being, what is special about such models. The quest for compositional RNNs is still on.

Host: Carlo Miniussi

CIMeC Talk

​When: Friday, 23rd February 2018, 3 - 4 P.M.

Where: ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Elena Nava - Psychology Department - Bicocca, University of Milan

​Title: Asymmetry signatures in the developmental population: evidence from space-number mapping and line bisection in infants and children

​Summary: Left-to-right number-space mapping is a fundamental trait of both human and non-human cognition, likely rooted in innate characteristics of the brain, as suggested by studies conducted in preverbal infants, humans in remote cultures and phylogenetically-divergent species. However, whether this number-space asymmetry persists or changes in early childhood has been little investigated, and the first part of the talk will focus precisely on this, in particular showing that the spatial representation of number is mapped onto external, adult-like frames of reference only by age 10, while it is still developing by age 6. Interestingly, a left-to-right asymmetry does not only emerge for numbers, but appears a default for any ordered sequence containing numerical, temporal, or verbal information. Thus, in the second part of the talk, I will propose that this left-to-right scanning may be the consequence of the earlier maturation of the right hemisphere, which drives allocation of visuo-spatial attention to the left hemispace. Indeed, preliminary findings in 4 month-old infants - whose eye movements were recorded during a baby-friendly version of the line bisection task - document a phenomenon of pseudoneglect that emerges very early in life.

Host: Giorgio Vallortigara

CLIC Seminar

​When: Wednesday, 14th February 2018, 11.30 A.M. - 1.00 PM

Where: 3rd-floor meeting room, CIMeC (Palazzo Fedrigotti), Corso Bettini, 31 Rovereto

Speaker: Jakub Szymanik (ILLC, University of Amsterdam) / http://www.jakubszymanik.com/

​Title: Learnability and Semantic Universals

​Summary: One of the great successes of the application of generalized quantifiers to natural language has been the ability to formulate robust semantic universals. When such a universal is attested, the question arises as to the source of the universal. In this paper, we explore the hypothesis that many semantic universals arise because expressions satisfying the universal are easier to learn than those that do not. While the idea that learnability explains universals is not new, explicit accounts of learning that can make good on this hypothesis are few and far between. We develop a model of learning — back-propagation through a recurrent neural network — which can make good on this promise. In particular, we discuss the universals of monotonicity, quantity, and conservativity and perform computational experiments of training such a network to learn to verify quantifiers. Our results are able to explain monotonicity and quantity quite well. We suggest that conservativity may have a different source than the other universals.

Contact: Sandro Pezzelle

CIMeC Talk

​When: Friday, 19th January 2018, 12.00 - 1.00 PM

Where: CIMeC Corso Bettini 31 Rovereto, 1st floor conference room

Speaker: Viviana Betti Department of Psychology, Sapienza University

​Title: Resting state networks as spatio-temporal priors for natural vision

​Summary: In the absence of a task, the cerebral cortex exhibits slow fluctuations of activity that are highly organized in large-scale spatio-temporal structures or resting state networks (RSNs). Resting state fluctuations are of great interest since they have been linked to task related activity patterns. A fundamental question today is the functional role of spontaneous activity, and what information, if any, is coded in these intrinsic patterns of correlated activity. In this talk, I will present the idea that the role of the intrinsic brain connectivity is to preserve and maintain an internal model of the environment that is built through the integration of information from visual and bodily inputs. Bodily inputs reflect the physical and the functional interaction that our body establishes with the external environment. In this vein, the hand has a special role because it is the primary means of interaction with the surrounding. This idea is based on recent studies using Magnetoencephalography (MEG) - a method with high temporal resolution and good spatial resolution. Here, we studied changes in static and dynamic functional connectivity, topology and integration measured with alpha and beta Band Limited Power (BLP) correlation at rest or during the observation of normal or time-scrambled movie clips. We showed that the audiovisual stimulation reduced the connectivity in alpha, reorganized the overall topology and changed the dynamics of nodal centrality. In contrast, the beta band intrinsic topology and dynamics of integration at rest were similar to those measured during the observation of normal movie sequences but altered for time-scrambled movies. Taken together, these findings suggest that the similarity between network connectivity at rest and during natural vision reflects an adaptation of the spontaneous activity to the naturalistic environment and common behaviors.

Host: Carlo Miniussi

PREMESOR Talk

​When: Tuesday, 9th January 2018, 9:30 - 10:30 AM

Where: ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Tommaso Pizzari - Edward Grey Institute, Department of Zoology, University of Oxford (UK)

​Title: Unravelling the cognitive basis of sexual behaviour in fowl

Host: Giorgio Vallortigara

PREMESOR Talk

​When: Monday, 8th January 2018, 11:30 AM - 12:30 PM

Where: ACN lab - seminar room, II floor, Piazza della Manifattura 1 - Building n. 14

Speaker: Diana Umeton, Centre for Behavior and Evolution - Institute of Neuroscience, Newcastle University

​Title: Unravelling the illusion of flicker fusion

​Summary: Pattern perception has been investigated in the context of predator-prey interactions for over 150 years. I will present a series of experiments investigating whether prey can deceive their predator through the visual illusion of flicker fusion. The flicker fusion effect occurs when high contrast repetitive patterns are perceived as blurred when prey are moving sufficiently quickly to exceed what predator eyes’ can temporally resolve. Although be an intriguing idea, the flicker fusion effect was so far poorly described and its camouflage efficacy has never being tested. For the first time, using praying mantis tracking computer-generated prey stimuli, we provide evidence in support of the flicker fusion effect and demonstrate that whether a pattern will be conspicuous or camouflaged depends not only on whether the object moves or not, but also on the speed at which it moves.

Host: Giorgio Vallortigara