Cogs2010 Laboratory introduction to models in Cognitive Science
Semester II, 2001

Overview

 

Cogs 2010 introduces basic principles and issues related to computational modeling of cognitive, perceptual and linguistics processes in cognitive science.  It introduces neural networks and other modeling frameworks, and emphasizes the significance of modeling techniques for cognitive science.

 

The subject includes weekly lab sessions, which provide an opportunity for students to familiarize themselves with some of the important models in the area.

 

Part 1: Neural Network models (weeks 1-6) includes neural network simulations

Part 2: Evolutionary computation (weeks 7-9) includes GA simulations

Part 3: Symbolic models (weeks 10) includes Copycat

Part 4: Embodied cognition (weeks 11-12)

Assumed Background

 

Students are assumed to have completed at least one unit  (4 credit points) of computer science, 1 unit of mathematics or statistics (such as MS, MT or PY103) and 6 units  of advanced subjects (second year or above) from subjects related to cognitive science (such as COMP, PSYC, PHIL, LING, IC, CS, EL, PD, PY or AN322). It is also recommended that students have completed COGS1000 (IC100) or COGS2000 (IC206) or an introductory artificial intelligence or cognitive science course.

Objectives

 

During the course students should:

 

·         acquire knowledge and understanding of connectionist and other computational models that are used to model cognition and intelligence.

·         gain an understanding of the cognitive phenomena to which these architectures apply.

·         gain an appreciation of fundamental cognitive issues that these architectures highlight.

·         acquire the skills to simulate these architectures.

Web Pages

 

Cogs 2010  Home Page: http://psy.uq.edu.au/CogPsych/cogs2010/ http://www.csee.uq.edu.au/~cogs2010

Connectionist Models of Cognition (CMC text book and simulator): http://www.csee.uq.edu.au/~cogs2010/cmc/

Cognitive Science Home Page: http://cognitivescience.uq.edu.au/

 

Lecturer Contact Details

 

A/Prof Janet Wiles

School of Psychology and School of Computer Science and Electrical Engineering

Room: 333, General Purpose South Building (bldg 78)

Email: j.wiles @ csee.uq.edu.au

Phone: 3365 2902

Office hours: Tuesday 2-4

Mode of Delivery: Times and Rooms

 

Thursday 4-6pm: Lecture in GP-South (bldg 78) room 344

Tuesday 4-6pm or Thursday 2-4pm: Laboratory sessions will be in GP-South, rooms TBA  Room 78-116.  Note that in the first week there will be no laboratory session.

Cogs2010 includes two different kinds of knowledge, content and skill.

·        ·        Content refers to the basic facts and ideas that underpin computational approaches to cognition. The content will be covered primarily in the lectures.

·        ·        Skill refers to the ability to construct, run, and interpret computational models using computer based simulation packages. Skill acquisition will occur primarily in the laboratory sessions.

Schedule

See below for the course plan.

See the online updated version here.

Textbook

The lab sessions will involve using several simulation packages.  The neural network exercises are from the online textbook “Connectionist Models of Cognition”. You can register online at http://www.connectionist.net/.

You don’t need to register to use Brainwave in the lab.  The local site for Brainwave is at

http://www.csee.uq.edu.au/~cogs2010/cmc/

 Hyperlinks to the appropriate chapters are provided in the schedule table. In order to access these chapters you need to register online first. The textbook contains interactive exercises using the BrainWave neural network simulator, which is built into the web pages. We recommend that you use Internet Explorer 4.0 or greater on Windows 95/98 for best results. To access the textbook outside of the laboratory times, you can use the library machines or use your own Internet connection.

Readings

Each week has assigned readings as listed in the Schedule table. Articles that are not online will be available from the UG course centre (on the second level of the Psychology Building). You are expected to read this material and the relevant simulation documentation prior to each lab. The readings in the table above and laboratory work are examinable.

Further readings are suggested at the end of each chapter of the online textbook. These readings are not compulsory but do provide background to the subject.

Schedule

 

Week

 

Starting

Lab Topic (Tuesday 4-6)

(Thursday 2-4)

GP-south 78-116

(Social Sciences Annex)

Lecture Topic (Thursday 4-6)

(Connell - 232) GP-south 78-344

Readings

1

2nd August

 

No lab

Introduction to Cognitive Modeling, neural networks and the BrainWave Simulator

Lecture by Scott Bolland

CMC
1 Preface;
2 Intro to NN;
3 Intro to BW

2

9th August

Introduction to BrainWave (ch 3)

The Interactive Activation and Competition Network and Jets and Sharks model (ch 4)

Lab Q&A.

Modeling Issues.

Automatic and controlled processing – Intro to lab - the Stroop effect.

 

CMC

4 IAC network

10 Stroop net

3

16th August

The Stroop model (ch 10)

Human memory modeling: making networks learn fast

 

Tba

4

23rd August

Memory

The Hebbian Network: The Distributed Representation of Facts (BrainWave)

 

Unsupervised learning: SOM

 

Tba

5

30th August 

The Self-Organizing Map: The Development of Feature Maps

(ch 8 Orientation map)

 

Supervised learning: backpropagation

Tba

6

6th Sept August

 

Backprop (ch 7) and Simple recurrent networks

Models of dynamically interacting agents: firefly synchronization and lego robot firefly demo

 

tba

7

13th Sept

Fireflies (BrainWave)

Models of evolutionary processes: introduction to evolutionary computation

 

Tba

8

20th Sept

EC I

Evolution of altruism, the role of game theory

 

Tba

mid-semester break

27th  September

 

Mid Semester Break

 

9

4th Oct

EC II

Evolution of human behaviour

 

Tba

10

11th October

EC III

Models of analogical reasoning: Copycat

Tba

11

18th October

Copycat (The CopyKitten simulator)

Embodied cognition: the role of robotics

tba

12

25th October

Reserved for Asgt work

Unified theories of cognition

tba

13

1st Nov

Reserved for Asgt work

Revision

 

Tba

 

Assessment

Oncourse assessment (60%): 10 weekly labs x 5% = 50%, 1 assignment worth 10%

Written answers to exercises are due at the end of each lab. Late penalty 1 mark per day unless given permission by the tutor. Lab exercises handed in later than 1 week will not be marked.

The assignment is due Thursday 1st November.

Examination (40%) – The exam will be held during the normal examination period and may include any material from the course content, including lectures, assigned readings and lab exercises. The exam will comprise short answer questions and worked problems.

Assessment Criteria for Grades

 

Laboratory Exercises: A student will have demonstrated a high level of performance in the lab exercises if in their submitted work they:-

·         Answer all of the procedural questions correctly.

·         Demonstrate a high level of understanding of how the architecture addresses relevant psychological phenomena and cognitive issues

·         Demonstrate some evidence of innovative thinking in the design or interpretation of a given model.

·         Submit all reports on time, in a neat and tidy form.

 

Examination: A student will have demonstrated excellent knowledge in the exam if their performance:-

·         Demonstrates understanding of the procedural details of all of the architectures covered in the class.

·         Shows they are capable of completing worked problems relating to these architectures.

·         Demonstrates an appreciation and understanding of how these architectures address relevant psychological phenomena and cognitive issues.

·         Demonstrates some evidence of innovative thinking in the interpretation of the models.

A grade of 7 will be awarded to a student who demonstrates excellent skills and knowledge as assessed by the examination, and who performs at a high level in the lab exercises and assignment.

A grade of 6 will be awarded to a student who demonstrates at least sound level of achievement/understanding in all assessment areas and achieves excellence in either the exam or the oncourse assessment.

A grade of 5 will be awarded to a student who demonstrates at least sound work in all assessment areas. Alternatively, a grade of 5 will be awarded to a student who demonstrates excellent knowledge/skill in one major area of assessment and basic knowledge/skills in the other areas.

A grade of 4 will be awarded to a student who demonstrates at least basic skills in all areas of assessment, or at least sound skills and sound knowledge levels in the majority of the assessed work.

Otherwise, a failing grade will be awarded

Plagiarism

Copying the assignment work of other students, allowing other students to copy assignment work by you and excessive collaboration are all regarded as plagiarism - a university offence. If you are in doubt about the permitted extent of collaboration, see the course coordinator.

When plagiarism is detected, no credit will be awarded for the component of assessment involved. If collaboration involves excessive cooperation, all parties involved will receive no credit for that component.

That having been said, discussion with colleagues (which should be duly acknowledged) is an essential part of the learning experience. But having learned material, the work submitted for assessment must be your own.

Please read the note included in the University's published literature on misconduct in the Student Handbook ... a survival guide.

Student Support Services

Any student with a disability who may require alternative academic arrangements in the subject is encouraged to seek advice at the commencement of the semester from a Disability Adviser at Student Support Services.