Dear All,
The CEU Department of Cognitive Science and the Center for Cognitive Computation (CCC) invites you to the following talk.
*Please note that this is an online event only, next Tuesday, starting from 3 PM.*
Speaker: Yanli Zhou (New York University)
Title: Compositional Learning of Function Interactions in Humans and Machines
Abstract: Humans are efficient learners of functions - the ability to represent and compose functions emerges early in development. Work in linguistics further suggests that humans are capable of learning much more complex function interactions. Going back to Kiparsky (1968), linguists have cataloged numerous linguistic phenomena covering four
logical patterns for ordering two interactive functions. “Feeding” in a context-free grammar is when one function creates the context for another to operate, and “counterfeeding” is the converse. “Bleeding” occurs when the operation of one function removes the context for another, and “counterbleeding” is its converse. In this project, our aim is to determine the extent to which humans and computational models can learn to compose functions based on the system of interactions. We introduce a learning task that adapts and extends the Piantadosi & Aslin (2016) design, evaluating participants on zero-shot function compositions covering all four interaction types. Our findings indicate that participants can correctly infer the underlying functions based on limited input-output examples; they can also generalize to novel combinations of functions across different conditions. Close examinations of participants’ response patterns reveal a number of potential inductive biases. Furthermore, we demonstrate that a standard sequence-to-sequence transformer model can be trained to complete the same task with high levels of accuracy via meta-learning. Incorporating guidance from human data, our model can learn to reproduce behavioral patterns that mirror the complete and complex way humans perform function compositions.
Time: 3 PM, Tuesday, April 16, 2024
Zoom: Meeting ID: 983 6360 8209<https://ceu-edu.zoom.us/j/98363608209?pwd=THFGZEFLVmVqaGlaMjlWYkxsa0dSUT09> Passcode: 951062
Please, be informed that video/photo recording might take place at the event and the edited version of the video material might be published to communicate or promote CEU PU's activities. Please, find our Privacy Notice here<https://www.ceu.edu/privacy>.
Best regards,
Ildikó Varga
Department Coordinator (Budapest)
Department of Cognitive Science
[cid:9cb0ecd0-e800-4392-a4d6-ba66545d05e7]
H-1051 Budapest
Nádor u. 15. FT room 404.
tel: +36-1 327-3000 2941
http://www.ceu.edu<http://www.ceu.edu/>
http://cognitivescience.ceu.edu<http://cognitivescience.ceu.edu/>
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu
Dear all,
We would like to invite you to the following talk by Richárd Reichardt, organized as part of the ELTE Cognitive Seminar series.
Time and date: 17:00 (CET), Thursday, 18 April 2024.
Speaker: Richárd Reichardt
Title: Is novelty memorable?
Abstract: One of the central questions of memory research pertains to the most important factors influencing memorability. Most experts consider novelty to be such a factor, even though empirical results are controversial on the memory effects of novelty. A possible solution for this conundrum might be based on the application of predictive coding principles to memory systems. This theory proposes a general information processing strategy characterizing the nervous system, which is based on predictions generated in accordance with prior experience that influences actual experience. From this standpoint, common sense novelty per se should not affect the updating of predictions or in other words memory encoding. However, if emphasis is placed on the degree of unexpectedness, we might paint a more consistent picture about one of the most important factors influencing memorability.
Zoom link: https://ppk-elte-hu.zoom.us/j/93539482032?pwd=cjBDcmlqOXZ5UHk0YUVRZ21XRWpTQ…
Meeting ID: 935 3948 2032
Passcode: 286094
If you have questions about the event, please contact us via email (kelemen.alexandra(a)ppk.elte.hu or reka.schvajda(a)ppk.elte.hu<mailto:reka.schvajda@ppk.elte.hu>).
We look forward to seeing you at the event,
Alexandra Kelemen
Réka Schvajda
organizers
ELTE Kognitív Pszichológiai Tanszék
Csaba Pléh www.plehcsaba.eu
vispleh@<http://@>ceu.edu Phone: 36303493735
Member HAS and Academia Europaae
CEU Dept of Cognitive Science Nádor utca 9 1051 Budapest, Hungary
Office: Nádor utca 9 4th Floor, Room 406
Dear all,
The CEU Department of Cognitive Science invites you to the following talk:
Alon Hafri, University of Delaware
Title: Where language and vision meet: Shared representational principles and content
Abstract:
There are few domains as central to cognitive science as language and visual perception. They are typically studied in isolation, yet at some level, they must connect. How? My talk will explore this connection. First, I reveal surprising parallels in how language and vision encode information: Vision, like language, can be compositional. I explore the psychophysics of visual scene composition, finding that the mind builds relational representations (e.g., a vase on a table) in a canonical order. Next, I show that language and vision tap into the same modality-neutral conceptual representations. Using symmetry as a case study, I find a striking correspondence between linguistic and perceptual judgments of the same visual displays. Lastly, I demonstrate how language can offer a window into non-linguistic visuospatial representations, leveraging the unique ways that certain languages encode such properties, particularly Mandarin Chinese. By examining how Mandarin speakers describe novel objects ("daxes") in different visual contexts, I uncover an unexplored influence of spatial relations (a dax in a bowl) on object shape computations. Taken together, my work highlights the representational principles and content shared by language and vision, with implications for how infants acquire language from visual observation.
Date: Wednesday, May 9, 2024
Time: 4 pm (to 5:30 pm) CET
Venue: Online, Zoom meeting 969 2496 5784<https://ceu-edu.zoom.us/j/96924965784?pwd=c2duZ0dDMFdEMUthK2Mwa2wzMllEUT09> (passcode: 471712)
Chair: Gergely Csibra
Best,
Bartu
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu
by DUCOG - Dubrovnik Conference on Cognitive Science
Dear All,
Registration is now open <https://ducog.cecog.eu/registration/registration>
for the XV. Dubrovnik Conference on Cognitive Science, which is
devoted to *Memory,
space and language**.* The conference will take place between 23 and 26 May
2024 in Dubrovnik, Croatia.
Register by 30 April 2024 to receive the early bird discounts.
DUCOG 2024 brings together researchers striving to understand what shared
neural and cognitive mechanisms allow humans and non-human animals to
represent and process memory, space, and language and how these mechanisms
change across the lifespan. Our goal is to uncover synergies and opposing
views of approaches from different levels of analysis, from cellular
through systems level neuroscience to cognitive- and neuropsychology, in
order to facilitate cross-talk between currently independent research
fields to inspire novel research.
*Invited speakers will include:*
Helen Barron – University of Oxford, UK
Melissa C. Duff - Vanderbilt University, USA
Paul Frankland – University of Toronto, Canada
Monika Schönauer - University of Freiburg, Germany
Jelena Sučević - University of Oxford, UK
For more information please visit https://ducog.cecog.eu
or email us at: ducog(a)cecog.eu
On behalf of the organisers,
Attila Keresztes,
Zsuzsanna Nemecz
- Conference chairs
Dear All,
The CEU Department of Cognitive Science and the Center for Cognitive Computation (CCC) invites you to the following talk.
*Please note that this event will be held at CEU Budapest site (N13. building), from 5 PM.*
Speaker: Ádám Gosztolai (EPFL)
Title: Interpretable representations of neural dynamics using geometric deep learning
https://arxiv.org/abs/2304.03376
Abstract: It is increasingly recognised that computations in the brain and artificial neural networks can be understood as outputs of a high-dimensional dynamical system conformed by the activity of large neural populations. Yet revealing the structure of the underpinning latent dynamical processes from data and interpreting their relevance in computational tasks remains a fundamental challenge. A prominent line of research has observed that task-relevant neural activity often takes place on low-dimensional smooth subspaces of the state space called neural manifolds. However, there is a lack of theoretical frameworks for the unsupervised representation of neural dynamics that are interpretable based on behavioural variables, comparable across systems, and decodable to behaviour with high accuracy.
To address these challenges, we introduce Manifold Representation Basis Learning (MARBLE), a fully unsupervised representation-learning framework for non-linear dynamical systems. Our approach combines empirical dynamical modelling and geometric deep learning to transform neural activations during a set of trials into statistical distributions of local flow fields (LFFs). Our central insight is that LFFs vary continuously over the neural manifold, allowing for unsupervised learning, and are preserved under different manifold embeddings, allowing the comparison of neural computations across networks and animals.
We show that MARBLE offers a well-defined similarity metric between different neural systems that is expressive enough to compare computations and detect fine-grained changes in dynamics due to task variables, e.g., decision thresholds and gain modulation. Being unsupervised, MARBLE is uniquely suited to biological discovery. Indeed, we show that it discovers more interpretable neural representations in several motor, navigation and cognitive tasks than generative models such as LFADS or (semi-)supervised models such as CEBRA. Intriguingly, this interpretability implies significantly higher decoding performance than state-of-the-art. Our results suggest that using the manifold structure yields a new class of algorithms with higher performance and the ability to assimilate data across experiments.
Time: 17:00, Wednesday, March 20., 2024.
Location: CEU Budapest site (1051 Budapest, Nádor u. 15.) N13. building, room 302.*
and Zoom (Meeting ID: 975 2152 9826<https://ceu-edu.zoom.us/j/97521529826?pwd=bXhOTDNzK054VFd2cUdMcVVCMkdUUT09> Passcode: 996748)
*Anyone not affiliated with CEU wishing to attend in-person in Budapest must RSVP to vargai(a)ceu.edu to get access to the lecture hall.
Please, be informed that video/photo recording might take place at the event and the edited version of the video material might be published to communicate or promote CEU's activities. Please, find our Privacy Notice here<https://www.ceu.edu/privacy>.
Best regards,
Ildikó Varga
Department Coordinator (Budapest)
Department of Cognitive Science
[cid:a9c19747-9116-484d-b203-0aeefb20295d]
H-1051 Budapest
Nádor u. 15. FT room 404.
tel: +36-1 327-3000 2941
http://www.ceu.edu<http://www.ceu.edu/>
http://cognitivescience.ceu.edu<http://cognitivescience.ceu.edu/>
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu
Dear All,
The CEU Department of Cognitive Science and the Center for Cognitive Computation (CCC) invites you to the upcoming event of the Budapest Computational Neuroscience Forum<https://ccc.ceu.edu/budapest-computational-neuroscience-forum>.
Speaker: Nikola Milićević, Pennsylvania State University
Title: Sensory systems and combinatorial neural codes
Abstract: Neural activity in sensory areas of the brain is shaped both by the stimulus and by the internal neural dynamics. When the stimulus space is known we can compute receptive fields of neurons. Receptive fields of individual neurons are convex in a number of brain regions (such as the hippocampus, and the visual cortex). The combinatorial neural code are the subsets of co-active neurons for some input to the neural network. Not any combinatorial code is compatible with convex receptive fields. This raises a natural question: how do recurrent networks produce convex codes? Towards this end, we study a recurrent neural network with the Dale’s law architecture.
We describe the combinatorics of equilibria and steady states of neurons in threshold-linear networks that satisfy Dale's law. The combinatorial code of a Dale network is characterized in terms of two conditions: (i) a condition on the network connectivity graph, and (ii) a spectral condition on the synaptic matrix. In the weak synaptic coupling regime, the combinatorial code depends only on the connectivity graph, and not on the synaptic strengths. Moreover, we prove that the combinatorial code of a weakly coupled network is a sublattice, and we provide a learning rule for encoding a sublattice in a weakly coupled excitatory network. Surprisingly, we find that the architecture of a Dale network “enforces” convex code output, in both strong and weak coupling regimes. Finally, we introduce a method inspired by game theory for inferring receptive fields, when the stimulus space is unknown or at least no consensus has been reached as in the case of olfactory systems.
Time: 17:00, Wednesday, 13 March, 2024.
Location: CEU Budapest (1051 Budapest, Nádor u. 15.) N15. room 104.
Zoom: Meeting ID: 931 3000 7576<https://ceu-edu.zoom.us/j/93130007576?pwd=cWdvcW5LWWx4blpyamxrSG5ySDFPQT09> Passcode: 142434
Should you have any inquiries about the series, please contact Mihály Bányai<mailto:mihaly.s.banyai@gmail.com>.
Please, be informed that video/photo recording might take place at the event and the edited version of the video material might be published to communicate or promote CEU's activities. Please, find our Privacy Notice here<https://www.ceu.edu/privacy>.
Best regards,
Ildikó Varga
Department Coordinator (Budapest)
Department of Cognitive Science
[cid:30304f38-4ad9-4ad6-9b95-ff6eafb9ebf3]
H-1051 Budapest
Nádor u. 15. FT room 404.
tel: +36-1 327-3000 2941
http://www.ceu.edu<http://www.ceu.edu/>
http://cognitivescience.ceu.edu<http://cognitivescience.ceu.edu/>
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu
Dear all,
The CEU Department of Cognitive Science invites you to the following talk:
Jean-Remy Hochmann (ISC Marc Jeannerod)
Title: Incomplete Language of Thought in infancy
Abstract:
The view that infants possess a full-fledged propositional Language of Thought (LoT) is appealing, providing a unifying account for infants’ precocious reasoning skills in several domains. In this talk, I'll present evidence that preverbal infants possess structured mental representations, with abstract generalizable content. However, I'll argue that careful appraisal of empirical evidence suggests that infants lack a crucial component of a propositional LoT: discrete representations of abstract relations.
Date: Wednesday, March 13, 2024
Time: 4 pm (to 5:30 pm) CET
Venue: D002* (QS Vienna) and Zoom (meeting ID: 969 2496 5784<https://ceu-edu.zoom.us/j/96924965784?pwd=c2duZ0dDMFdEMUthK2Mwa2wzMllEUT09>, passcode: 471712)
Chair: Erno Teglas
If you want to schedule a meeting with Jean-Remy, please indicate your availability here<https://doodle.com/meeting/participate/id/dN7vMApe>!
Best,
Bartu
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu
Dear all,
The CEU Department of Cognitive Science invites you to the following talk:
Manuel Bohn (Max Plank Institute/Leuphana University Lüneburg)
Grounding socio-communicative development in everyday experience: an individual differences perspective
Social cognition and communication are defining aspects of what it means to be human. In this talk, I want to present our research program that studies the driving forces behind socio-communicative development. Our goal is to understand how everyday social interactions influence development. To get there, we will have to overcome numerous theoretical and methodological hurdles. I will present the steps we have taken in recent years towards this goal, which include computational models of the cognitive processes underlying aspects of socio-communicative development, tasks to measure cognitive abilities on an individual level, methods to capture and quantify everyday experiences and international collaborations to probe the generalisability of findings. Much of our work is still in the early phases, so I am very much looking forward to feedback.
Date: Wednesday, March 6, 2024
Time: 4 pm (to 5:30 pm) CET
Venue: D002* (QS Vienna) and Zoom (meeting ID: 969 2496 5784<https://ceu-edu.zoom.us/j/96924965784?pwd=c2duZ0dDMFdEMUthK2Mwa2wzMllEUT09>, passcode: 471712)
Chair: Gergley Csibra
*Anyone not affiliated with CEU wishing to attend in-person in Vienna must RSVP here<https://forms.office.com/pages/responsepage.aspx?id=E1nE2VN24kuSC72wOGOBhAH…> to get access to the lecture hall.
If you want to schedule a meeting with Manuel, please indicate your availability here<https://doodle.com/meeting/participate/id/avmBr5ge>!
Best,
Bartu
______________________________________________
Subscribe by sending an empty mail to talks-subscribe(a)cogsci.ceu.edu
Unsubscribe by sending an empty mail to talks-unsubscribe(a)cogsci.ceu.edu