Scribd
Upload
3 views15 pages

Computation and Cognition - 4

The document discusses various aspects of decision-making, including reward representations, computational modeling, and the anatomy of decision processes. It highlights the complexity of choices, the role of context in valuation, and the neural evidence for common currency coding in decision-making. Additionally, it covers methodologies in empirical research and brain imaging techniques used to study cognitive processes.

Uploaded by

Krishna Prasad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
3 views15 pages

Computation and Cognition - 4

The document discusses various aspects of decision-making, including reward representations, computational modeling, and the anatomy of decision processes. It highlights the complexity of choices, the role of context in valuation, and the neural evidence for common currency coding in decision-making. Additionally, it covers methodologies in empirical research and brain imaging techniques used to study cognitive processes.

Uploaded by

Krishna Prasad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15

Sequence representations

Post
Ant

Miyapuram, 2004 Hikosaka et al., 1999 Bapi et al., 2006


Reward representations

Perception Imagination
Reward

No Reward

Dopaminergic midbrain

Midbrain activation (cluster average)


Visual presentation  CS+  CS-
Support Vector Machine Classification Imagination  CS+  CS-
Associative Representations
Parahippocampal gyrus

CS US

Smell

Experience
reward imagination
Feel
medial orbitofrontal cortex

Taste
Perception Imagination

Holland (1990) Salivate


UR
reward perception
Computational Modelling
• Hebbian learning: Associative strength increases with co-occurrence of a pair of
stimuli (contiguity)
• Expected Utility: Associative strength directly related to (conditional) probability
of co-occurrence of a pair of stimuli (contingency)
• Reinforcement learning: Associative strength directly related to discrepancy
between expected and actual outcome (prediction-error principle)
• Markov-model: The present outcome is dependent on previous n outcomes
Anatomy of a Decision
A simple choice is more complex than we think!

1. Valuation of options
Predicted valuation signals

Learning
2. Choice
Action valuation signals

3. Outcome
Experienced valuation signals

5
Perceptual Decision Making
LIP Responses scale with stimulus difficulty

Roitman & Shadlen (2002)


Sequential Sampling Models
Dynamic model of choice that evolves through evidence
accumulation within a trial

Philiastides and Heekeren (2009)

Drift Diffusion model takes into account both Accuracy and RT distribution to break-
down decision process into psychological (model) parameters – drift rate (v),
threshold (a), bias (z), and non-decision time (Ter)
7
Sequential Sampling models

8
Valuation in Decision-Making
• What computations are carried out in the brain to make different types
of decisions?
• How are these computations implemented?

Value
Apple Action Action Action
Sweetness 2
= 1 3
Value
Sweetness
+
Calories Value
Calories
Max
+
Value
Acidity
Chosen
+ Action
Acidity
…
9
Context Dependence
• The valuation of options depends on the context of other
available options
• Violates the assumption of rational-choice theories that
“Decisions reflect absolute valuations assigned to individual
options”

10
Common Currency Coding in Decision-Making
• What neural evidence exists for ‘common currency’ coding across
domains of decision-making?
• No single study, No single paradigm
• Overwhelming number of fMRI studies!!
• Variability of findings across participants, experiments, scanner strength, analysis
techniques
• Consistency: identified through Meta analysis
• Specificity: large-scale Meta analysis (NeuroSynth)

Perceptual: Value: Individual Social: Herd Behavior,


Ambiguous Sensory Preferences and Game Theory
Information Liking

11
11
General approach

Modelling Research
Reinforcement Learning, Machine
Learning, Artificial & Natural Neural
Networks, Brain Reading,…

Cognition Theoretical Research


Empirical Research
Behavioural & Neural experiments, Top-down & Bottom-up processes,
Cognitive processes & their interactions Cognitive interactions, e.g. learning
e.g. learning, attention,memory, & decision making, …
multidisciplinary approaches,…

Methodological Research
Laboratory setup,Collaborations,
Meta analysis, Literature, Topic identification, …
Brain Imaging
Magnetoencephalography
(MEG)

Efflux
–
Example:
Influx

Woldorff et al (1993)
Mismatch
Negativity +

Magnetic Field Voltage Distribution


The skull is transparent to magnetism -- less blurring
Deep and radial dipoles are invisible from outside the head
Computational Models
Reinforcement Learning

Brain Computation Behavior


How is it What is Why is it
Computed? computed? computed?

You might also like