iClicker questions
Check
- the semantic question (15)
- The priming question (17)
- Attention neglect (20)
What leads a neuron to fire, signaling that it has detected relevant input?
a) Most neurons receive direct input from the world, so they fire when they see (or hear, or feel, etc.) things in the world that they are tuned to detect.
b) Neurons receive a mix of excitatory and inhibitory inputs, and they fire if the excitatory inputs win the competition and get them over threshold to fire.
c) Neurons receive language-like inputs, so a neuron will fire if the neurons it receives input from send relevant instructions (e.g., “It’s election day so I should vote.”).
b - Most neurons are direct sensory neurons, in brain helping processing. They also do not receive language inputs. Instead, they receive excitatory and inhibitory inputs.
Which of the following is NOT true of neural networks?
a) Bidirectional connections support pattern completion (taking partial information and filling in the rest). Think faces sim.
b) Feedforward inhibition anticipates incoming activity levels, and feedback inhibition reacts to activity levels, together regulating overall activity. Think inhib sim.
c) Attractor dynamics prevent networks from settling into stable states. Think necker cube sim.
d) Categories/similarity can be captured by overlapping distributed representations. Think cats-and-dogs sim.
c - They help them settle into stable states, not prevent.
Which of the following is NOT true of learning in neural networks?
a) Learning is driven by patterns of neural firing.
b) Effects of learning are captured in the strength of connection weights (synapses) between units.
c) Self-organizing learning picks up on statistical patterns; error-driven learning generates responses that match correct targets.
d) Learning cannot solve problems that are harder than simple one-to-one mappings.
d
Which of the following contribute to the formation of topographically organized edge detectors in V1? (select all that apply)
a) lateral excitatory connections among V1 neurons
b) the commonality of edges in the world
c) ocular dominance columns
d) self-organizing learning that picks up on regularities in the world
e) the optic chiasm
a) Lateral connections
b) Commonality of edges in the world
d) self organizing learning
Still not
How does the V1 receptive field simulation form edge detectors?
A) They are built in from the start, to simulate genetic programming
B) They develop through self-organizing learning that picks up on the regularities of edges in the world
C) They develop through error-driven learning that requires tuning in to edges to identify objects correctly
D) They develop through reinforcement learning, with greater than expected rewards when units detect edges
b
How does the object AA model solve the computationally challenging task of recognizing objects despite dramatic variations in retinal input images for a given object (e.g., based on the position or size of the object)?
a. Across layers, the network learns to represent increasingly complex combinations of features that are also increasingly spatially invariant
b. The network is reinforced for deep learning across layers via lateral excitatory connections
c. The network has built-in feature detectors across layers that map retinal inputs across different positions and sizes to the appropriate object representations at the highest level of processing
d. The network develops activation-based receptive fields that show hierarchical invariance
a
Parkinson’s patients have reduced levels of dopamine. Dopaminergic medication increases dopamine and “fills in the dips.” Why does such medication cause some Parkinson’s patients to develop gambling habits? The medication:
a) reduces all probabilistic learning, so patients have difficulty generalizing to new situations.
b) shifts the balance of Go and NoGo learning, so that wins count for more, and failures count for less.
c) leads to less learning about positive outcomes, so that patients seek larger positive outcomes through gambling.
d) reduces pattern separation in the hippocampus, so that episodic memories of losses blend with episodic memories of wins.
b
Dopamine dips:
a) increase
b) decrease
c) do not affect
- activity of Go pathway neurons in the striatum
- the strength of associated Go pathway corticostriatal synapses
- activity of No Go pathway neurons in the striatum
- the strength of associated No Go pathway corticostriatal synapses
- b
- b
- a
- a
Dopamine bursts:
a) increase
b) decrease
c) do not affect
- activity of Go pathway neurons in the striatum
- the strength of associated Go pathway corticostriatal synapses
- activity of No Go pathway neurons in the striatum
- the strength of associated No Go pathway corticostriatal synapses
- a
- a
- b
- b
Why does prefrontal damage (or a less developed prefrontal cortex) affect behavior especially under conditions of conflict? Because conflict is when:
a) error-driven learning is least likely to be helpful
b) spatially invariant representations require dopamine-driven learning
c) pattern completion (via attractor dynamics) is essential for ensuring that goals are kept separate rather than becoming integrated
d) sustained neuronal firing is needed to bias processing in a way that is different from what the system would tend to do otherwise
d