Dynamates Symposium

Emanating from the Dynamates consortium, this symposium brings together experts in the field of cognitive neuroscience, cognitive biology and computational modelling to address key questions in predictive auditory processing with a focus on spatial and temporal auditory perception in volatile environments. It will cover purely behavioural approaches as well as combinations with eyetracking, EEG and other neuroimaging modalities. 

Speakers:
- Florent Meyniel (NeuroSpin, CEA Paris-Saclay & Neuromodulation institute, Sainte Anne Hospital, Paris, France)
- John van Opstal (Donders Centre for Neuroscience, Netherlands)
- Barbara Shinn-Cunningham (Carnegie Mellon Neuroscience Institute, USA)

Date:
Monday, June 24, 2024 (as a satellite event of the FENS Forum 2024)

Venue:
Campus of the University of Vienna
Talks in the "Alte Kapelle"
Posters in the Neue "Aula"
You may use one of the access points to the Campus at Spitalgasse, 1090 Wien
Map locating the two buildings "Alte Kapelle" and the "Aula"

Maximum number of attendees:
80 (first come first serve)

Registration:
- Form: https://leidenuniv.eu.qualtrics.com/jfe/form/SV_22WteXETC6Ay47A
- Deadline: March 27, 2024 (23:59 CET)
- Registrants encouraged to submit a poster abstract 
- Participation fee: 50.- € (lunch and coffee breaks included) 
- Optional evening social at Zum Gangl (extra 30.- €)
- Accommodation not included

Program:

09:00 - 09:30 Get together

09:30 - 10:30 Florent Meyniel
Hierarchical probabilistic inference for accurate prediction
Accurate predictions are crucial for perception and decision making in dynamic and noisy environments. This talk makes the claim that the human brain spontaneously utilizes local statistical regularities to predict the next item in a sequence. The type of statistical regularities that the brain utilizes is currently debated, I propose that it reflects the sequential structure of the world and relies, at a minimum, on transition probabilities between consecutive items. Interestingly, the brain leverages local statistical regularities even when they emerge from stochastic fluctuations, leading in this case to erroneous predictions. Why does the brain sometimes rely more than it should on the recent past? One possibility is that the brain is adapted to a world that is changing, where the recent past is more informative about future observations than the distant past. Evidence shows that the human brain can detect changes in sequence structure by making a nearly optimal use of the available evidence. These findings indicate that hierarchical probabilistic inference underlies auditory predictions in the human brain. A key notable aspect of this probabilistic account is the association of predictions with confidence levels, indicating their precision and likelihood of accuracy. The talk demonstrates that humans can introspectively access this confidence, which also modulates neural responses triggered by unexpected sounds. Artificial neural networks trained to predict the next item in a sequence develop a representation of this confidence that regulates their inference process. These results highlight the importance of incorporating confidence representations for accurate predictions in dynamic and noisy environments. This requirement aligns with probabilistic inference, can be implemented in simple neural networks, and appears to be operative by the human brain.

10:30 - 11:00 Coffee break

11:00 - 12:00 Barbara Shinn-Cunningham
How task goals and stimuli affect recruitment of cognitive networks 
In healthy, normal-hearing listeners, prefrontal cortex manages complex negotiations between cognitive networks involved in setting task goals and focusing and storing task-relevant information. This talk will review fMRI, EEG, and fNIRS experiments exploring how task goals and task difficult modulate prefrontal cortical activity. These different modalities all demonstrate that spatial auditory tasks lead to very different patterns of recruitment in prefrontal cortex than nonspatial tasks.

12:00 - 13:00 Lunch break

13:00 - 15:00 Poster session

15:00 - 15:30 Coffee break

15:30 - 16:30 John van Opstal
Using Reaction Times to Characterise the Auditory System of Normal and Impaired Human Hearing, and of Non-Human Primates
Although reaction times have since long been a valuable tool for studying the neural mechanisms underlying sensorimotor integration in the visuomotor field, they have so far hardly been applied in studies of the auditory system. I will show that the behavioural reaction times evoked by sounds are very sensitive to subtle acoustic manipulations, and that they can be used to uncover different aspects of neural processing within the auditory system. A reaction time is the combined result of (i) sensory processing at the peripheral system, (ii) perception, selection, and decision making within the central nervous system, and (iii) motor planning and execution. In general, the harder it is to extract relevant stimulus information from the (noisy) sensory input, the longer the reaction time. They are easy to measure, require little instruction to the participant, and are hardly confounded by higher cognitive mechanisms. The so-called LATER model (Carpenter & Williams, Nature, 1995), provides a simple, yet powerful, neuro-computational account for the underlying statistics of reaction-time distributions that can be used to analyse auditory-evoked data. In my talk, I will show how reaction times can be used to reliably measure the spectral-temporal sensitivity of the auditory system of normal-hearing listeners, macaque monkeys, and cochlear-implant users. In our recently started project OtoControl-2.0, we will use this approach with the aim to improve on current CI fitting strategies.

16:30 - 17:30 TBA

19:00 - 22:00 Dinner/social (optional)