Summary Concepts in HMS - Coordination

Coordination – Lieke Peper
General introduction
You have to coordinate the muscle activity and the body; you can have high
muscle strength but the coordination is important as well
You have to interact with other muscles, the environment (objects, other people,
animals)

We pick up information with our senses (vision, sound, smell, touch)
We get information about the environment and about our relation with the
environment

Sensorimotor loop
(perception-action coupling)
- Nervous system can generate muscle activity
- Information about the environment is needed for the right movement
- Information picked up by our senses
- Also generate new sources of information
o Proprioception (of our own body)
o Through other senses
- Coupling between what you do and the perceptual consequences  re-
afference
- Coupling between generating movement and resulting sensory
information  sensorimotor loop/perception-action loop

Information
Information
Information is based on sensory signals that enter the CNS by our senses and
gives us information about the environment, our relation with the environment
and about the status of our body

How is information obtained from sensory stimulation
Various perspectives
Different definitions of “information”

Computational approach
Nervous system is doing some processing with the signals to obtain
the information

Bayesian information processing
- There is always some noise influencing our perception
- Uncertainty in sensory signals
- With prior information  reduction of uncertainty
o Expectations, previous experience or opponent told
you
- Nervous system combines the prior (expectation) with
likelihood (actual information) and this gives you the
posterior (final perception)
o Posterior provides an estimation of the location
where the ball will land and the amount of
uncertainty is smaller than for the prior and likelihood (reduction of
uncertainty)

,Prior
- Probability that the value is ‘x’
- Based on prior experience/knowledge
- The prior can be more tight, but in this example there
is more uncertainty in the prior than in the likelihood
- Depends on experience/knowledge

Likelihood
- Probability of observation ‘o’ if the value is ‘x’
- For something you are observing there is a peak at a given
value (most likely observation) and some uncertainty
- Reflects the degree of perceptual uncertainty
- Peak  most likely perception that you have


Posterior
- Probability that the value is ‘x’, given observation ‘o’
- = prior x likelihood (normalized to probability of observation ‘o’)
- The uncertainty in the posterior is always reduced
- Don’t need to know the equations in the graphs
- Final perception that you have

The combination of the 2 probability distributions yields a better estimate
- More information  additional likelihood  more reduce of the uncertainty
- Works for other types of perceptions and combination of perceptions (not
only visual)

How does the peak/width of posterior relates
- Stronger peak of likelihood  peak of posterior is closer to 2
- Width is smaller because you have more certainty because of more
information

Perceptual uncertainty
- Reflected by likelihood distribution
- Prior distribution is not affected
o Because of bad weather  you know the roads, so the prior is
always the same, the way you see it because of the weather is seen
in the likelihood
o More sensory noise (perceptual uncertainty) does not affect the
prior
- Increasing perceptual uncertainty
o Posterior:
 Lower peak for likelihood, prior not affected  lower posterior
 More uncertainty is lower peak and wider distribution
 More uncertain
 Relative importance of likelihood ↓
 Relative importance of prior ↑ (more reliance on prior in bad
weather)
- The higher the likelihood, the more important for the posterior
- The weight of prior increases when there is more perceptual uncertainty
o It results from the combination of the probability distributions, there
is no active weighting in the brain

, Ames room
- Prior is affected by the room
- The experience is different from what you expect
- You have to look through the peep hole to get this perception
- The illusion occurs because the prior affects your perception




Recap Bayesian perspective
- Sensory perception is characterized by uncertainty
- Perception results from information processing
o Aimed at reducing uncertainty
o Involving statistical representations of perceptual uncertainty
(likelihood) and prior experience/knowledge (prior)
- Information: reduction of (perceptual) uncertainty

Ecological perspective
Information processing
- Ecological approach: natural perception is ‘direct’  no information
processing necessary
They say that the Ames room does not really help to study core principles of
visual perception
- Natural perception is not through a peephole (like the Ames room)
- Ambiguities resolved by changing point of observation
- Information in the ‘optic array’


Michaels & Carello (1981)
‘indirect’ perception
- Processes needed to interpret stimuli
Direct perception  in general, no information processing/computation is
required because information is already present in the structure of the light that
surrounds you: the optic array

“no interpretation needed, because information is available in the ‘optic array’”
(J.J. Gibson)

Optic array: structured, ambient light
- The light that falls into your eye
- The light is structured because the light is
reflecting the environment.
- Information about the environment is reflected by
the light and that can enter your eye
- If the perceptual system is able to pick up these
structures, than it can directly perceive what is around you, and you don’t
need any interpretations
- You can see texture elements or reflection patterns
- Tells you something about your orientation/relation to the objects

People who are watching the same scene, and stand next to each other, are their
optical arrays identical?
- No, because their points of observation are slightly different