Planning in Action Dynamical Modeling of Developing Decision-making Processes Ralf Cox

1
Planning in ActionDynamical Modeling of
Developing Decision-making ProcessesRalf Cox
Ad Smitsman

• Behavioral Science Institute
• Dynamical Systems Group

2
Program
Behavioral Science Institute
Planning in Action

• Hand use (limb selection) Basic results
  Mechanisms
• Planning (of hand use)
• Perseveration experiment - Adults
• Model
• Perseveration experiment - Young children
• Conclusions and Prospects

3
Hand Use in Adults - Borrowed
Behavioral Science Institute
Planning in Action
Set up by Gabbard, Misaki Rabb, 1997 Gabbard
Rabb, 2000, 2004
4
Mechanisms of Hand Use - Borrowed
Behavioral Science Institute
Planning in Action

• Two mechanisms (working next to each other)
• Motor dominance ? manifested in handedness
• Attentional information ? spatial or task-related
• Motor dominance is controlling factor in
  programming hand use at midline and ipsilateral
  hemispace.
• Attentional information alters the programming of
  the movement in contraleteral hemispace.
• ? Cannot really explain the frequency
  distributions.
• ? As yet no model combines these two mechanisms.

5
About Making Choices
Behavioral Science Institute
Planning in Action

• Planning is a decision-making process for actions
  to take (or inferences to make) with a number of
  defining properties. It is
• a dynamical process
• a multi-causal process
• an embodied process
• an embedded process

• But also
• subjected to preferences
• working on several time-scales (history)

6
The experiment we performed is as follows
Behavioral Science Institute
Planning in Action

• Adult participants 14 righthanders and
  10 lefthanders
• Three training conditions
• Random (Gabbard)
• Non-preferred hand (test 1)
• Preferred hand (test 2)

• Test 4 training (T) trials followed by 2
  neutral (N) trials
• T-trials are on (-)10o and (-)30o
• N-trials are on (-)90o

7
Hand-use Perseveration Results (preliminary)
Behavioral Science Institute
Planning in Action

• In the Random condition we replicated Gabbard et
  al.s results.
• In the two Test conditions

8
Dynamical Model for Hand Use
Behavioral Science Institute
Planning in Action

• Lateral activation function u(t) for each hand
  (site)
• Obeying a continuous dynamics
• Inhibitory coupling between the sites
• Neurobiological noise
• Motor dominance Direction and strength as
  difference in resting levels between the
  sites. ? Larger and opposite in righthanders
  compared to lefthanders.
• Attentional information Asymmetrical input to
  the sites.

9
Simulations of Hand-use Perseveration
Behavioral Science Institute
Planning in Action
With this model we can reproduce (simulate) the
frequency distributions of adult hand use in the
Gabbard experiment.
 

• A second (internal) input source is added to the
  sites
• ? Build up of a memory trace during the training
  trials.

10
Planning in Young Childrens Tool Use
Behavioral Science Institute
Planning in Action

• Tool use involves decision making for action on
  several levels and time-scales.
• Earlier decisions prepare later ones.
• Decision-making process is continuous and under
  constant influence of
• various perceptual input
• (motor) preferences
• training or memory of earlier actions

Experiment Model Effect of training and hand
preference on the decision which hand to use for
grasping a spoon.
11
Hand-use Perseveration - Experiment
Behavioral Science Institute
Planning in Action

• Children of 14 (N21), 24 (N19) and 36
  (N26) months old
• Three conditions
• Hand-preference condition
• Test condition
• Control condition

12
Hand-use Perseveration - Results
Behavioral Science Institute
Planning in Action

• 24 36 mo significant perseveration
• 14 mo not significant

• subgroup significant perseveration

13
Simulations
Behavioral Science Institute
Planning in Action

• Resting-level difference is age dependent
• ? Largest in 36 mo and smallest in 14 mo.

Examples
14
Conclusions
Behavioral Science Institute
Planning in Action

• Hand use
• The decision which hand to use is affected by
  previous decisions.
• This decision is mediated by the strength of the
  preference.

• The model realistically
• combines perceptual input, motor preference and
  memory
• simulates behavior as task and context dependent
• reproduces individual and group differences
• Future work on the model
• Simulate timing effects in the task.
• Adjustment of hand choice (switch) during the
  task, due to changes in the perceptual-motor
  workspace (earlier studies).