Computational Cognitive Neuroscience
Master Program in Cognitive Science, Comenius University in Bratislava

Time/Place: Summer Semester 2018/2019, room I8, Lectures: Thursday 14:00 - 15:30, Labs: Thursday 15:40 - 17:10

Credits: 6

Lecturer: Prof. Ľubica Beňušková
Teaching assistant (TA): Mgr. Ing. Matúš Tuna - For individual project consultations and/or project deadline extensions plz email to: tunamatus@gmail.com

Course aims

Computational cognitive neuroscience relies upon theories of cognitive science coupled with neuroscience and computational modeling. In this course, we will study neurobiological processes that underlie cognition by means of computational models. We will address the questions of how cognitive processes are affected and controlled by neural circuits in the brain. Modeling of some basic mechanisms of cognitive functions will be done using the Emergent simulator.

Assessment

Project (guidelines) - due 23 May 2019 20%
Lecture Presentation (guidelines) 20%
Lab Exploration in Emergent (guidelines) 20%
Final exam - 1st term 30 May, 2nd term 6 June 2019 at 14:00 40%

Marking

0-60 % Fx
61-65 % E
66-72 % D
73-80 % C
81-90 % B
91-100 % A

Course schedule

Date Lecture Presentation
Student name / paper
Exercises in Emergent [ Student] Reading
21.02. Course evaluation. Introduction to computational cognitive neuroscience. Main concepts in modelling. (L1 slides) The first contact with the Emergent simulator O'Reilly,ch.1; Farkas (2012)
28.02. Spiking neurons models. Biology of individual neuron and its implementation in Emergent. (L2 slides) Single neuron
Neuron as a detector
O'Reilly,ch.2
07.03. Structure of cortical networks, localist and distributed representations, excitation and inhibion of neurons. (L3 slides) Lauren / Visual competition
Inhibition
Necker Cube
O'Reilly,ch.3
14.03. Biological mechanism of memory and learning, long-term potentiation and depression of synaptic efficacy.
(L4 slides)
Adam / Synaptic metaplasticity
Pattern Associator
Cats and Dogs
O'Reilly,ch.4 (up to XCAL)
21.03. Self-organization, error-driven learning, combination of both. (L5 slides) Laura / Memory retention
Self-organisation
Error-driven Learning
O'Reilly,ch.4 (XCAL)
28.03. Functional organisation of the brain. Overview of brain areas. (L6 slides) Hansoo / Functions of hemispheres
Family Trees
Face Categorization [ Viera ]
O'Reilly,ch.5
04.04 Visual perception, attention, bottom-up and top-down mechanisms. Spatial neglect. (L7 slides) Lucia / Attention
V1 receptive fields
Invariant Object Recognition
O'Reilly,ch.6 Ungerleider & Pessoa (2008) Ward (2008)
11.04. Motor control and reinforcement learning. (L8 slides) Viera / Motor control
Action selection [ Lucia ]
Reinforcement Learning
O'Reilly,ch.7
18.04. Maundy Thursday - public holiday Easter No reading
25.04. Learning and Memory -- Semantic and episodic memory, implicit vs. declarative, priming, familiarity, etc. (L9 slides) Barbora / Cortex and memory Hippocampus*[ Laura ]
ABAC Task [ Barbora ]
Long-term priming [ Peter ]
Short-term priming [ Hansoo ]
O'Reilly,ch.8
02.05. Language. (L10 slides) Danijela / Vocabulary learning
Dyslexia [ Danijela ]
Semantics [ Rebecca ]
Sentence Gestalt [ Lauren ]
O'Reilly,ch.9
09.05. Executive functions, the role of frontal cortex. (L11 slides) Peter / PFC control
Stroop effect [ Matija ]
A not B [ Adam ]
O'Reilly,ch.10; Rougier et al. (2005); O'Reilly (2010)
16.05. Agency, theory of mind, self-awareness. (L12 slides) Matija / The sense of agency
Rebecca / Observing self
Postponed presentations

Recommended literature

Labs

Emergent Simulator

Install the simulator according to the instructions on this page.

If you are using Slovak language in your operating system, to be able to open the Excercise Projects in Emergent you have to set your delimiter to . instead of , in the regional settings.

The Emergent software is still in development. Be patient. If you encounter any issues opening the Exercise Project or additional files within the project in Emergent (such as pre-trained weights) do not hesitate to contact the teaching assistant.

Useful links

Assesment details

Project Spiking Neurons - guidelines

The Simple Model of Spiking Neurons of Izhikevich (2003) holds a special position among the variety of spiking models proposed, namely because it is simple yet allowing to simulate a large spectrum of firing patterns of biological neurons. The task in this assignment is to study the original paper, explore various neuron firings and summarize your findings in a report. The report in PDF format is to be sent to the teaching assistant (TA) through email by the deadline. You can write you report either in English or Slovak. The choice of programming language is up to you, it is allowed to help yourself with the source code in the paper or other source. The emphasis of this project is on how you display the results and how you describe and explain the results in terms of clarity and quality of presentation. This assignment has to be completed individually, and be prepared to demonstrate your working source codes to the TA (on demand). The assignments has 2 parts.

Part A: Firing Patterns (10%)
Implement the model (or use available codes) to vary parameters to achieve 8 firing patterns (RS, IB, CH, FS, TCa, Tcb, RZ, LTS) from Izhikevich (2003) (Fig. 2). Display the firing patterns in graphs and comment on them briefly. If you choose to implement the model yourself, implement differential equations of form dx/dt = f(x) using the forward Euler method: x(t+h) = x(t) + hf(x(t)), where h is the step size (try 0.5 or 1 millisecond). More help can be found on the author's webpage under Research and Models of spiking neurons.

Part B: Network of Neurons (10%)
Implement (or use available codes) of a neural network of 1000 spiking neurons with random connections as described in part IV of Izhikevich (2003) paper. Show the network's ability to self-organize causing the neurons to fire synchronously in time as in Fig. 3. of the paper. Explore, display in pictures and comment on (a) the influence of the inhibition-excitation strengths ratio and (b) the influence of thalamic noise magnitude on the model's behavior. Answer the following questions: Under which conditions (i.e. values of which parameters) does the network show periodic synchronizations? Under which conditions the spiking of neurons does NOT synchronize at all? Any other intersting behavior you can observe?

Examples of reports: 2013 Dinga R., 2012 Lajtoš M., 2011 Blanář J., 2010 Budiš J., 2009 Malinovský B.

Presentation (Lectures) - guidelines

Presentation will be based on the scientific paper you receive from the lecturer (and of course any other valuable sources you yourself obtain). The presentation should take up about 20 minutes (plus 5 minutes for questions and discussion) and will be delivered after the lecture. Here are some useful GUIDELINES how to make a good paper presentation (replace 'you' with 'they' and do not forget the title slide!). You should sign up for a concrete date and topic (Lectures column of Schedule) through email to both lecturer and TA.

Exploration in Emergent (Labs) - guidelines

The goal of this assignment is to give a presentation in the Labs part of the class on a specific topic you choose from Excercises column above. Instructions related to each Exploration in Emergent are linked within the page. At the beginning of your presentation, you should explain the particular phenomenon to your classmates using few slides. Then the demonstration of the Emergent exploration follows. The whole presentation should last around 20 minutes (plus 5 minutes for questions and discussion). You should sign up for a concrete date and topic (Labs column of Schedule) through email to both the TA and lecturer.

NOTE: The Emergent software is still in development. You may encounter issues opening the Exercise Project or additional files within the project in Emergent (such as pre-trained weights). Due to this fact we advise you to allocate enought time to prepare for your presentation of Exploration in Emergent before the actual presentation date.