~Chapter 5 - Lecture Section 5.3

Question 1

Any object or scene is going to undergo ___.

Answer 1

Multiple stages of analysis

Question 2

Objects and Scenes activate ___.

Answer 2

multiple modules

Question 3

Objects and Scenes are ___.

Answer 3

multi-dimensional

Question 4

True or False: Different dimensions of an object are often processed by different regions

Answer 4

True

Question 5

For the perception of faces, the start of Cortical processing starts in the ___

Answer 5

V1

Question 6

For the perception of faces, what does the Fusiform Face Area (FFA) do?

Answer 6

The FFA is the specialized face identity Module

Question 7

For the perception of faces, what does the Amygdala do?

Answer 7

The Amygdala identifies emotional aspects of the face (e.g. ☺)

Question 8

For the perception of faces, what does the Superior Temporal Sulcus (STS) do?

Answer 8

Gaze direction and speech movements

Question 9

For the perception of faces, what does the Frontal Lobe do?

Answer 9

The Frontal Lobe analyzes Attractiveness

Question 10

What is the purpose of a Binocular Rivalry Task?

Answer 10

To try to explicitly link neural activity to object perception

Question 11

What are the steps in a Binocular Rivalry Task?

Answer 11

The left and right eye are shown different pictures, and the subjects use buttons to indicate which image they are seeing whilst having an fMRI done. Instead of seeing a melded-up combination of both pictures, what happens is the input from one eye becomes suppressed, so it’s not perceived, and the person perceives the other picture.

Question 12

What do the eyes do during a Binocular Rivalry Task?

Answer 12

After some amount of time, the eye spontaneously switches to the other, nobody can explain why the switching happens, but it is a well-known and repeatable phenomenon.

Question 13

During the Binocular Rivalry Task when subjects reported seeing a face, where was more activity detected?

Answer 13

The Fusiform Face Area (FFA)

Question 14

During the Binocular Rivalry Task when subjects reported seeing a house, where was more activity detected?

Answer 14

The Parahippocampal Place Area (PPA)

Question 15

What did the Binocular Rivalry Task done on monkeys in the Infratemporal Cortex (IT) indicate?

Answer 15

It indicated that there was a very close association between activity in the IT and the monkeys’ perception

Question 16

In the Binocular Rivalry Task done on monkeys, activity in ___ was not well correlated with the monkey’s perceptual report.

Answer 16

lower areas (e.g. V1)

Question 17

During the Binocular Rivalry task done on Monkeys, when the monkey is reporting seeing the starburst, the butterfly neuron had ___ in its firing

Answer 17

no change

Question 18

At a later time during the Binocular Rivalry task done on Monkeys, when they report seeing the butterfly, the butterfly encoding neuron now fires ___, so there was significant changes in the IT butterfly neurons activity, which were accompanied by subsequent change in the monkey perceptual report

Answer 18

robustly

Question 19

Perception in animals and human subjects is more closely related to activity in ___ visual areas, and less related to activity in ___ visual areas

Answer 19

higher-order // lower-order

Question 20

What is the purpose of Brain-Reading technology?

Answer 20

Brain-reading technology is related to object perception in that different modules in different regions in the brain will be activated by different stimuli, and so when you look at a particular object, there will be a pattern of activation

Question 21

How does Brain-Reading technology work?

Answer 21

Instead of giving insight into how the brain actually produces perception of an object, brain-reading technology treats the activation pattern in your brain when looking at an object as a pattern recognition task

Question 22

How is a Decoder created?

Answer 22

There are computer algorithms that receive fMRI imaging information and they use this information to create a Decoder

Question 23

In fMRI, each chunk of brain that is activated is referred to as a ___.

Answer 23

Voxel

Question 24

Each different image could create a ___ pattern of ___ over the visual areas

Answer 24

different // Voxels

Question 25

How is the Decoder trained?

Answer 25

Training the Decoder involves knowing what the stimulus is, measuring the Voxel activation to the stimulus, and sending both the stimulus identity as well as the Voxel pattern to a Decoder, so it trains this Decoder knowing what the stimulus, and what the pattern of activation is.

Question 26

How can you test a Decoder?

Answer 26

By giving it only the pattern of brain activation, and then asking it to predict what stimulus is shown.

You then compare this prediction with what the subject was actually shown.

Question 27

In early work, Decoders were used to predict the ___ of very simple stimuli.

Answer 27

orientation

Question 28

Decoding what a person is looking at can be considered a ___ task.

Answer 28

pattern recognition

Question 29

What did Jack Gallants lab develop?

Answer 29

A more complex brain reader that is able to process more naturalistic images, and the performance of this has been improved by incorporating into the Decoder, a modeling of spiking activity of neurons that would lead to the observed bold responses.

Improved by modeling the spiking activity of neurons that may lead to the observed BOLD responses

Question 30

How did Gallants brain-reader work?

Answer 30

Had subjects sit in an fMRI machine, and showed them millions of images that were taken from Youtube, and they created two kinds of model.

They trained the Decoder on half the images, and then tested the Decoder on the other half of the images

Question 31

What was the Structural Encoding Model in Gallants brain-reader?

Answer 31

The Structural Encoding Model was about the structure and features of each image. So it should have light and dark parts in similar places, texture in similar places…

Question 32

What was the Semantic Model in Gallants brain-reader?

Answer 32

The Semantic Model included tagging the semantic features of each scene. E.g. Does it have a building, does it have wildlife?

Question 33

In Gallants brain-reader, the performance is ___ when you combine the Structural and Semantic Encoding Models

Answer 33

better