Developing a model for location based route learning in a virtual world
B ACHELOR THESIS , BY M ARCEL Z UUR
A
UGUST11, 2013
U
NIVERSITY OFT
WENTEF
ACULTY OFB
EHAVIORALS
CIENCES, P
SYCHOLOGYD
EPARTMENT OFC
OGNITIVEP
SYCHOLOGY& E
RGONOMICSF
IRST SUPERVISOR: P
ROF.
DR. F
RANK VAN DERV
ELDES
ECOND SUPERVISOR:
DR. M
ATTHIJSN
OORDZIJ1
Abstract
A robot has been given a route learning task. It’s goal is decision making based on the recognition of situations. It features a behavioural model which includes object recognition based on the ventral stream, dual-process decision making and motor control. This model tries to follow the computational cognitive neuroscience (CCN) ideals. Implementation is done using a combination of neural networks and programming.
PCA reveals that representations can emerge at different level of processing. Lesions study and PCA shows that location detection can be enhanced by combining vision and sonar. Results also show the benefits from using dual-processing decision making.
This thesis ends with stating that combining CCN modelling with traditional research can provide a powerful tool in understanding cognition.
Contents
1 Introduction 4
2 Cognitive Computational Neuroscience 7
2.1 Historical context . . . . 7
3 The robot, it’s environment and the route learning task 12 3.1 The robot and it’s environment . . . . 12
3.2 Route learning task . . . . 15
3.3 Skills needed to complete the route learning task . . . . 15
3.4 Research question . . . . 16
4 Model for object recognition, decision making and motor control 17 4.1 Ventral pathway for object recognition . . . . 17
4.2 Decision making: a dual process approach . . . . 21
4.3 Motor control . . . . 23
5 Implementing the model 24 5.1 Materials . . . . 24
5.2 Neural Networks . . . . 25
5.3 Programming . . . . 32
6 Training 37 6.1 Stage 1 learning objects and system 2 activation . . . . 37
6.2 Stage 2 training for action decision . . . . 39
6.3 Stage 3 testing the neural networks and the decision making model . . . . 42
6.4 Stage 4 training locations and sequence . . . . 43
2
Contents 3
7 Analysis 48
7.1 Producing output . . . . 48
7.2 Analysis of the sonar detection network . . . . 50
7.3 Analysis of the location network . . . . 54
7.4 Sequence results . . . . 61
7.5 Activation of system 2 . . . . 66
8 Conclusion and Discussion 68 8.1 PCA Analysis . . . . 68
8.2 Lesion Studies . . . . 69
8.3 Model . . . . 69
8.4 Discussion . . . . 71
Bibliography 76 A Leabra training algorithm 82 B Programming code 84 B.1 Detected objects . . . . 84
B.2 Creating the robot and the virtual world . . . . 85
B.3 Model implementation . . . . 85
1
Introduction
Reductionist biology–examining individual brain parts, neural circuits and molecules–has brought us a long way, but it alone cannot explain the workings of the human brain, an information processor in our skull that is perhaps unparalleled anywhere in the universe. We must construct as well as reduce and build as well as dissect
(Markram, 2012).