Read on Twitter
A thread:
None of the robotics labs on campus are taking new PhD students, so it looks like my third rotation will be in a decision making neuroscience lab.
As such, I turned to the literature to get an ecological psychology perspective on decision making. I started 1/n
None of the robotics labs on campus are taking new PhD students, so it looks like my third rotation will be in a decision making neuroscience lab.
As such, I turned to the literature to get an ecological psychology perspective on decision making. I started 1/n
with this paper by Araújo et al.:
http://eprints.qut.edu.au/4816/1/4816_1.pdf
on the ecological dynamics of decision making in sports contexts. The paper led me to think more deeply about a few questions that have been bugging me for some time: 2/n
http://eprints.qut.edu.au/4816/1/4816_1.pdf
on the ecological dynamics of decision making in sports contexts. The paper led me to think more deeply about a few questions that have been bugging me for some time: 2/n
1. What are “goals” and “plans” from a non-representational / non-cognitivist perspective? What is the relationship between goals and plans?
2. How should we think about the role of the nervous system in so-called “higher cognition” phenomena like decision making?
3/n
2. How should we think about the role of the nervous system in so-called “higher cognition” phenomena like decision making?
3/n
Araújo et al. make the case that we call decision making emerges dynamically in the interaction between an agent and the environment under multiple constraints.
Interactions between multiple constraints govern the maintenance and transition between action patterns. 4/n
Interactions between multiple constraints govern the maintenance and transition between action patterns. 4/n
The ability to make decisions requires being sensitive to context dependent ecological information. In the framework of Warren’s behavioral dynamics, a decision is made as the agent-environment dynamical system relaxes to an attractor.
To choose between different decisions 5/n
To choose between different decisions 5/n
is to relax to one attractor over others. Which attractor (decision) is chosen will depend on the initial conditions of the agent-environment system and on the basin of attraction for each attractor.
Of course, which decisions are even available will depend on the 6/n
Of course, which decisions are even available will depend on the 6/n
structure of the vector field, i.e. what affordances exist and whether or not the agent is sensitive to the information specifying these affordances.
The kind of information an agent is sensitive to will depend that agent’s developmental and learning history. 7/n
The kind of information an agent is sensitive to will depend that agent’s developmental and learning history. 7/n
Goals play a crucial role in this framework. As Araújo et al. say,
“... to make a decision is to direct a course of personal interactions with the environment towards a goal.”
This is in line with Warren’s use of the word “goal” to describe the attractors in 8/n
“... to make a decision is to direct a course of personal interactions with the environment towards a goal.”
This is in line with Warren’s use of the word “goal” to describe the attractors in 8/n
the behavioral dynamics and the previous discussion of decision making as falling into an attractor state.
As @__stillPoint has pointed out in reference to this article by Donella Meadows,
http://donellameadows.org/archives/leverage-points-places-to-intervene-in-a-system/
this gives goals an odd causal status, one that is 9/n
As @__stillPoint has pointed out in reference to this article by Donella Meadows,
http://donellameadows.org/archives/leverage-points-places-to-intervene-in-a-system/
this gives goals an odd causal status, one that is 9/n
outside of the organism so that we can no longer speak of an organism as “having a goal”. Generally, I think this is a good way to think about goals in a non-representational way. But it is not the only way.
We can also turn toward Mark Bickhard’s discussion of goals. As 10/n
We can also turn toward Mark Bickhard’s discussion of goals. As 10/n
Bickhard says,
“...goals... need only be internal functional switches that, for example, switch back into a trial and error interactive process or to a learning process under some conditions (functional failure), and switch to further processing in the system under other 11/n
“...goals... need only be internal functional switches that, for example, switch back into a trial and error interactive process or to a learning process under some conditions (functional failure), and switch to further processing in the system under other 11/n
conditions (functional success)... This stands in strong contrast to common conceptions of goals as representing environmental goal conditions.” - Bickhard and Terveen, 1995, p. 62.
While this conception of goals brings concerns about what the right observer independent 12/n
While this conception of goals brings concerns about what the right observer independent 12/n
notion of “function” is, it is another non-representational way of thinking about goals that allows us to place goals internal to the organism.
These two conceptions of “goals” are fundamentally different, the first referring to outcomes particular to the task dynamics and 13/n
These two conceptions of “goals” are fundamentally different, the first referring to outcomes particular to the task dynamics and 13/n
the second referring to state dependent switches in the organism. Both are valid but play different explanatory roles.
Plans also play a role (though not a central one in the eco psych framework) in decision making. As noted by Araújo et al.,
“...plans are just one way of 14/n
Plans also play a role (though not a central one in the eco psych framework) in decision making. As noted by Araújo et al.,
“...plans are just one way of 14/n
constraining actions...”
But they do not make it clear how we should conceive of plans in an ecological sense.
The standard way of conceiving of plans is as representations of action sequences or sequences of motor commands. How should we think about plans in a 15/n
But they do not make it clear how we should conceive of plans in an ecological sense.
The standard way of conceiving of plans is as representations of action sequences or sequences of motor commands. How should we think about plans in a 15/n
non-representational way? And what is their relationship to goals?
In light of the preceding discussion, what we call plans might actually be an emergent property of different contexts modulating which affordances we are sensitive to. Alternatively, we could think of 16/n
In light of the preceding discussion, what we call plans might actually be an emergent property of different contexts modulating which affordances we are sensitive to. Alternatively, we could think of 16/n
plans from Stepp and Turvey’s anticipatory systems framework
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2827858/
as anticipatory sub-systems (non-representational or maybe representational in the interactivist sense) that constrain future perception-action possibilities.
As for how plans relate to 17/n
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2827858/
as anticipatory sub-systems (non-representational or maybe representational in the interactivist sense) that constrain future perception-action possibilities.
As for how plans relate to 17/n
goals, I’m still not too sure, but we could perhaps think of goals as providing contexts within which plans emerge. This could be the sense in which a plan is “towards a goal”.
The above discussion offers some intuitions for how to think about the role of the nervous 18/n
The above discussion offers some intuitions for how to think about the role of the nervous 18/n
system in decision making. I don’t have concrete ideas, but the modulatory and selectionist roles of the nervous system as discussed in these two great papers on Gibsonian neuroscience by @JBruineberg and @PAdrianFrazier
https://www.tandfonline.com/doi/full/10.1080/10407413.2019.1615204
https://www.tandfonline.com/doi/abs/10.1080/10407413.2019.1615205
19/n
https://www.tandfonline.com/doi/full/10.1080/10407413.2019.1615204
https://www.tandfonline.com/doi/abs/10.1080/10407413.2019.1615205
19/n
can help in constraining and inspiring theories about the role of the nervous system in decision making. 20/20
