PS3060: Perception and Action
Term II, MONDAY 10 – 12 am (Room 128 Wolfson)

Lecture 1: Introduction, Basics of Sensory-Motor Control (Orientation and Navigation)

Course co-ordinator: Johannes M. Zanker, j.zanker@rhul.ac.uk, (Room 218)


Topics Lecture 1:


The Problem

Everyday activities like chopping a carrot, driving a car, or playing the piano require precise coordination of sensory and motor systems, produced by our brain continuously and apparently without any particular effort. Inappropriate behavioural control, e.g. in road accidents or in neurological disorders like Parkinson’s disease, demonstrate what kind of, and what amount of information processing is involved in even the simplest activities.

Fundamental aspects of sensory-motor coordination are illustrated by a number of examples from biology, cognitive psychology, neurology and robotics.
The goals of the course are:

The program and organisational/formal details can be found at : http://www.pc.rhul.ac.uk/ug-dir/courses/PS306.html

Details about the coursework can be found at : http://psyserver.pc.rhbnc.ac.uk/zanker/teach/Ps3060/Ps3060_poster.htm


Perception-action loops: the control of behaviour

What information does the brain process, when you make a cup of tea?

click here for some suggestions what task the brain has to solve during this comparatively simple sequence of activities…

note in all of these action components there is a close, bi-directional relationship between
sensory input and motor commands !!!!

Looking at the eye movements of someone making a cup of tea can reveal some of the sensory aspects of these sensory-motor coordination sequence. (Land MF, Mennie N, Rusted J, 1999)


Stabilisation, Orientation and Navigation

a few simple examples of complex but effortless (and mostly unnoticed) brain activities in the coordination of sensory and motor information:


stabilisation 
  •  sailors (and most passengers) stay upright on the boat in strong winds
  • songbirds don't fall of the swinging twig of a bush,
  • hummingbirds hover in mid-air


         orientation

  • humans can easily walk along a corridor, or even keep a car on the road
  • bees approach flowers as well as artificial sugar sources
  • bacteria swim towards food sources (nutrient gradients)
  • protista grow towards the light
  •  lobsters navigate in geomagnetic maps (Alerstam 2003)



navigation
  • you found room 128 this morning, ...
  • ... and will find your way home!
  • whales travel around the world to find specific breeding or feeding waters

note that different biological systems face the same problems:
do they come up with similar solutions?

you will find many stunning examples of orientation behaviour from the animal kingdom in Waterman (1989).


Feedback, open and closed loop systems

sensory input affects behaviour - behaviour changes sensory input : sensory-motor loop

such loops are formalised in control theory (systems theory); instead of comprehensive mathematical descriptions, which precisely quantify such relationships, we here will only use illustrations in terms of circuit diagrams.

cybernetics: analysing sensory input and feedback connections and using the fundamental control principles to design ‘autonomous’ machines that automatically stabilise, orient, navigate, carry out tasks (one recent goal: autonomous helicopter, click here

a simple example - feedback loop for an ocular following reflex :
(much the same circuit could be used to illustrate the control of your heating by means of a thermostat)

if feedback is prevented experimentally (or not used in a particular action component, for instance in saccades), we speak about an open loop (or ballistic) system, as opposed to a closed loop system wich includes feedback about the consequences of the motor action

please note that the reality is less simple: there are many different types of eye movement (motor patterns), interactions with head, body movements, variations in dynamics (the timing aspects of a control system), and several sensory inputs (networks instead of loops)

you can find a bit more about feedback control systems in Houk & Lehman (1987)


Vehicles

the most fundamental principles of sensory-motor control can be incorporated in ‘vehicles’ (Braitenberg, 1984)

the particular, control logic predicts behaviour
a vehicle comprises three basic steps:


the simplest possible vehicle has only one sensor, one connection and one motor: its angular orientation is fully determined by external disturbances - it generates a random walk that only can be modulated in its speed, which nevertheless can be used to change the probability of staying inside/outside the area close to a stimulus!

more behavioural variation, in particular directed movements, are possible, when a vehicle is constructed with two sensors (to measure changes of a stimulus) and two motors (allowing controlled rotations)

option 1: sensory stimulation increases motor output: excitation

differential effects of the stimulus on the left/right vehicle motor will lead to changes in speed and direction, depending on the distance from and angle towards the stimulus!


crossed positive connections lead to contralateral excitation >> the vehicle turns towards the stimulus because the closer sensor, which is stimulated more strongly, will increase the thrust on the other side (and vice versa); the vehicle will get faster because detected stimulus intensity, and the magnitude of the motor output, increases when the vehicle gets closer >> ‘aggression’

uncrossed positive connections lead to ipsilateral excitation >> the vehicle turns away from the stimulus because the closer sensor, which is stimulated more strongly, will increase the thrust on the same side (and vice versa); the vehicle will initially get faster because detected stimulus intensity grows, and therefore the magnitude of the motor output increases when the vehicle gets closer; it finally will sit and rest in some distance >> ‘fear’

option 2: sensory stimulation reduces motor output: inhibition

once again, differential effects of the stimulus on the left/right vehicle motor will lead to changes in speed and direction, depending on the distance from and angle towards the stimulus!



uncrossed negative connections lead to ipsilateral inhibition >> the vehicle turns towards the stimulus because the closer sensor, which is stimulated more strongly, will reduce the thrust on the same side (and vice versa); the vehicle will slow down because detected stimulus intensity grows, and therefore the overall motor output is reeduced, speed will diminish when the vehicle gets closer >> ‘blind love’

crossed negative connections lead to contralateral inhibition >> the vehicle turns away from the stimulus because the closer sensor, which is stimulated more strongly, will reduce the thrust on the other side (and vice versa); the vehicle will initially slow down because the overall detected stimulus intensity grows, and therefore the motor output is increased, and in turning away will speed up again and move on >> ‘attentive admiration’

general orientation behaviour : chemotaxis, phonotaxis, phototaxis can be easily explained by such mechanisms

larger sets of sensors & connections can be employed for generating a ‘value system’, more complex vehicles can be designed by additional sensors, action systems, processing strategies

for some moderately complex examples, including groups of vehicles, you can go to:   http://www.cogs.susx.ac.uk/users/christ/popbugs/intro.html

on the other hand, these vehicles (sometimes also called turtles) have found their way into the world of toys, after the robotics community discovered LEGO mindstorms as simple simulation tools, see: http://el.media.mit.edu/logo-foundation/workshops/roboprojects.html


key reading:

some study questions

download lecture handout


back to index page
last update 23/02/2004
Johannes M. Zanker