100% satisfaction guarantee Immediately available after payment Both online and in PDF No strings attached
logo-home
Lectures. AI for an open society $8.68   Add to cart

Class notes

Lectures. AI for an open society

1 review
 65 views  9 purchases
  • Course
  • Institution

this paper includes all six lectures of AI for an open society , plus article Elliot et al.

Preview 4 out of 43  pages

  • June 23, 2022
  • 43
  • 2021/2022
  • Class notes
  • Leendert van maanen
  • All classes

1  review

review-writer-avatar

By: ece125345 • 6 months ago

avatar-seller
AI for an open society

Lecture 1: intro

Die topics met uitleg nog ff opzoeken

Lecture 2 Models of perceptual decision making
- Allan Newell and Herbert Simon: personal scientific heroes
- The basic building block: A cognitive model of evidence accumulation
- How can this cognitive model be useful: Effect of time pressure
- Another implementation: Diffusion Decision Model
- Outlook: New directions in User Models

- Herbert Simon
 Dartmouth workshop on artificial intelligence
 He tried to understand how businesses work by looking at the decision making
processes of those businesses and of those individuals that work for those
businesses. And the method of choice is bodies of cognition and so forth.
- Allen Newell
 Timescales of human action; what he explains on this page and the pages around
is that when you talk about behavior/decisions, things that people do, you can
talk about this on various timescales. So if you’re on the top, you can tak about
this in terms of decisions that are made over the course of months or weeks. And
really at the bottom, you can talk about micro decisions that are made inside the
brain, the biological events.
-Biological band; lowest, not even cognition, more biology. How do you do the
neurons in your brain send signals to other neurons in your brain. This is at the
very lowest level of how behavior comes about.
-Cognitive band; in 1 sec you have a reaction time distribution. These are very
simple decisions based on perceptual things you see that lead to decision
processes on a time scale of about a second. You can model that using these
kinds of processes
-The rational band; processes that occur on a slightly higher level and also take a
little bit more time. An example is thinking of reasoning about causal
probabilities: Today is a little bit sunny, better than yesterday, and we can reason
about what the probability is that it is sunny, given that I know, for example, that
I had an ice cream and I had a sunburn (because these two are events that occur
because it is sunny, causal relationship).
- The social band; models that illustrate or help you understand why certain
individuals take a particular political position. Time course of weeks, maybe
months.

,Focus now on cognitive band:
- Actions/decisions that are performed in the order of seconds
- Sometimes referred to as cognitive operations. Perception, memory, attention
- Example: a sniper, when he decides to shoot or not, first has to think about what it is
he or she is looking at
- You can see all these tasks as consisting of many of these cognitive operations.
- A small building block of cognition
- (Die test met twee gele stippen 1 kleine rode en dan witte bolletjes bewegend
eromheen, welke kant bewegen de witte stipjes op?) The evidence accumulation
calculus model assumes that what you did was that you were kind of weighting the
information that you saw for the left foot motion or the rightward motion. So,ass you
collected evidence for a leftward motion and or evidence for a rightward motion until
you just had enough evidence. This model (onder) assumes that you have an average
drift and on every talk or every choice that you make you can be slightly above the
average or slightly below the average and what you are is drawing from a normal
distribution.

, Sometimes there is a slight difference in this start point distribution (bijv. Met die
stipjes test kan je denken dat het naar rechts gaat doordat docent rechts staat), so
sometimes you start a little bit higher, so closer to the finish line, the threshold than
other times.




- So, two assumptions you make; on every trial, the drift is a value from this normal
distribution. The evidence accumulation is slightly faster or slower, and on every trial
the start point is slightly different;




- The final step; These are accumulators and every option is believed to have it’s own
accumulate. If you combine all those assumptions, you end up with a model called
- The fastest accumulator is the winner, so that is the choice you’re going to execute
- Model fitting: you try to find the optimal set of parameters
What are the optimal parameters?

,  Startpoint; there’s not much variability in where the choices start. So, if you
always consider all options the same, this upper bound should be zero.
 Threshold; means how much evidence do I think I need in order to make a
decision that’s high? You need to sample for a long duration. That also means
that if the process takes longer, if it’s low, then you’re going to make a quick
decision
 Drift rate; the speed in which you accumulate evidence
 Non-decision time; you can think of this as the time you need to press the button
(Over experiment dat je elke keer op een knop moet drukken als je de stimulus
ziet)
- If you choose different values of drift rate, you predict different shapes of direction
time distribution
- Example: effect of target location in 2AFC
???????
- We made a couple of assumptions.
One is that the drift rate, so the speed of evidence accumulation, is a function of the
angle between the two alternatives. So, the lower the angle, the lower also the
duration. You can think of this as; it is more difficult to extract information and The
other assumption was that the threshold functions as a variance, is a function of the
instruction. So you can imagine that if you tell participants to be accurate they just
think about things longer. So, threshold values are higher for accuracy than for
speed. But speed more errors.
When you lower the threshold, you choose A instead of B. Then you need less steps
to reach that threshold.

The benefits of buying summaries with Stuvia:

Guaranteed quality through customer reviews

Guaranteed quality through customer reviews

Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.

Quick and easy check-out

Quick and easy check-out

You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.

Focus on what matters

Focus on what matters

Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!

Frequently asked questions

What do I get when I buy this document?

You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.

Satisfaction guarantee: how does it work?

Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.

Who am I buying these notes from?

Stuvia is a marketplace, so you are not buying this document from us, but from seller juliahaverbeke. Stuvia facilitates payment to the seller.

Will I be stuck with a subscription?

No, you only buy these notes for $8.68. You're not tied to anything after your purchase.

Can Stuvia be trusted?

4.6 stars on Google & Trustpilot (+1000 reviews)

72042 documents were sold in the last 30 days

Founded in 2010, the go-to place to buy study notes for 14 years now

Start selling
$8.68  9x  sold
  • (1)
  Add to cart