final lists

Question 1

PP objections

Answer 1

1. dark room problem
2. mathematical formalization
3. not a neurophysiological theory
4. falsifiability

Question 2

four neural predictions for PP (Walsh paper)

Answer 2

1. error-signalling neural responses to sensory stimuli should scale inversely with expectation
2. top-down signals represent sensory prediction
3. at each level of the cortical hierarchy there are two functionally distinct subpopulations representing predictions and prediction errors
4. prediction error minimization is achieved through repciprocal exchange of error and prediction signals across levels - a process known as ‘hierarchical inference’

Question 3

evidence enacted perception

Answer 3

• perception is constituted by sensorimotor knowledge

1. experiential blindness
2. no perception when images are stabilized on the retina
3. self-movement is necessary for perception to develop
4. neuroscience studies support inside-out view

Question 4

cognitivism → enactive view

Answer 4

• cognition is understood as computations over mental representations
• the subject of cognition is a detached observer with a bird’s eye view of the world
• intentionality is explained by the representational nature of mental states
• the architecture of cognitive systems is conceived as being highly modular
• processing in subsystems is assumed to be largely context-invariant
• computations are considered neutral to the body or environment (functionalism)
• models of cognition take into account only the inner states of a cognitive system, and ignores how the body or environment might affect thinking (individualism)

Question 5

cognitivism → enactive view

Answer 5

• cognition is understood as capacity of generating structure by action
• the cognitive agent is immersed in his/her task domain
• system states acquire meaning by virtue of their role in the context of action
• the functioning of cognitive systems is thought to be inseparable from embodiment
• a holistic view of the architecture of cognitive systems prevails, which emphasizes the dynamic nature and context-sensitivity of processing
• models of cognition take into account the embedded and ‘extended’ nature of cognitive systems

Question 6

4E cognition

Answer 6

1. embodied: cognition is not isolated in the brain but the body’s structures and processes shape how we think
2. embedded: cognition is embedded in a relevant external context
3. enacted: actions help shape our cognitive processes, meaning cognition is partly constructed through the things we do, not just the things we think
4. extended: cognition can be extended beyond the brain and body, into the environment

Question 7

dynamic directives

Answer 7

dispositions for action, embodied in dynamic patterns of activity

Question 8

pragmatic implications of enactive view

Answer 8

1. study the relation between neural activity patterns and action contexts
2. avoid studying passive subjects but use paradigms that involve active exploration

Question 9

barsalou: in-between view of representations

Answer 9

1. there is no permanent static representation of hammers in long-term memory, built from amodal symbols, that is loaded into WM identically across different occasions
2. the representation that does reside in long-term memory results from superimposed effects of associative learning distributed across relevant sensorimotor systems, with the resultant network changing constantly after every learning episode (and overlapping considerably with networks for other categories)
3. when this distributed network is accessed, it produces one of infinitely many hammer representations dynamically
4. these representations serve temporary representational functions by providing useful inferences in specific situations

Question 10

barsalou: cognition mediates between sensation and action

Answer 10

1. cognition can represent non-present states: this implies dynamic, multimodal representations that are temporarily constructed in WM to exand our action repertoire
2. cognition can create conceptualizations through language: allows humans to adapt behavior in novel ways by categorizing and appraising experiences

Question 11

cognitive enhancement: current directions for future research

Answer 11

1. take individual differences in training effectiveness into account
2. use realistic training approaches
3. more research testing transfer of training to real life WM performance

Question 12

cognitive enhancement: critical design features

Answer 12

1. active control group
2. multiple transfer tasks
3. multiple training tasks
4. follow-up assessments
5. daily life measurements
6. large sample size
7. preregistration
8. double blind design

Question 13

hype cycle phases

Answer 13

1. technology trigger:a potential technology breakthrough kicks things off. early proof-of-concept stories and media interest trigger significant publicity. often no usable products exist and commercial viability is unproven
2. peak of inflated expectations: early publicity produces a number of success stories — often accompanied by scores of failures. some companies take action; most don’t
3. through of disillusionment: interest wanes as experiments and implementations fail to deliver. producers of the technology shake out or fail. investment continues only if the surviving providers improve their products to the satisfaction of early adopters
4. slope of enlightenment: more instances of how the technology can benefit the enterprise start to crystallize and become more widely understood. second- and third- generation products appear from technology providers. more enterprises fund pilots; conservative companies remain cautious
5. plateau of productivity: mainstream adoption starts to take off. criteria for assessing provider viability are more clearly defines. the technology’s broad market applicability and relevance are clearly paying off

Question 14

traditional view of visual WM vs current view

Answer 14

• traditional: WM is just for holding visual information because otherwise this information would disappear (i.e., has a fixed nature)
• current: WM is for guiding action, facilitated through action, and as a result of moving around

Question 15

visual WM is for action

Answer 15

1. objects in WM can be (re)prioritized to help guide current and future action
2. markers of visuospatial and motor-plan selection occur concurrently

Question 16

visual WM occurs through action

Answer 16

1. control of the overt gaze and covert attention are linked
2. the oculomotor system controls internal attention

Question 17

visual WM is a consequence of action

Answer 17

1. we look at item locations and its spatial mapping to preserve multiple spatial frames
2. we hold information about past events and make future predictions to hold moving objects in wm (account for shifting spatial frames)

Question 18

improve deep learning systems

Answer 18

1. embedding hard-code rules in DNNs (top-down learning)
2. more variable training in richer 3D environments (interacting with and learning from environment)
3. learning from less data (learn from prior experience)
4. supplementing basic pattern-matching with reasoning abilities (create their own algorithms through reasoning)

• also, active inference in AI models

Question 19

fallacies about AGI

Answer 19

1. narrow intelligence is on a continuum with general intelligence (scaling up leads to better task specific skills but doesnt generalize)
2. easy things are easy and hard things are hard (moravecs paradox)
3. the lure of wishful mnemonics (can lead to misunderstandings or exaggerated expectations)
4. intelligence is in the brain (its actually embodied)

Question 20

Chollet: why LLMs are not intelligent

Answer 20

1. they memorize, this is only a part of intelligence
2. they cant solve previously unseen problems
3. they fail the ARC benchmark

Question 21

Chollet: necessary behavior that would show AGI

Answer 21

1. the ability to synthesize new templates or adapt new ones to solve novel problems
2. this would mean they pass the ARC benchmark

Question 22

Chollet: how to overcome LLMs intrinsic limitations

Answer 22

1. allow them to do on-the-fly tuning to learn and adapt to new situations as they come

• this is essentially active inference

Question 23

Phylogenetic refinement

Answer 23

behavioral and cognitive theories are progressively elaborated based on the evolutionary sequence of changes in organisms over time.

Question 24

how does phylogenic refinement make us understand brain function

Answer 24

• behavior through the lens of interactions with the environment
• explain brain systems as elaborations of ancestral control mechanisms
• by tracing the evolutionary development of simpler behaviors into more complex systems in modern species, with the goal of generating biologically plausible models that reflect the actual mechanisms evolved over millions of years

Question 25

phylogenetic refinement: control of interaction

Answer 25

The brain manages interactions with the environment through feedback loops, adjusting actions (like foraging or escaping) to maintain balance in the organism’s stat

Question 26

phylogenetic control: nested feedback loop

Answer 26

Brain function is built on multiple layers of feedback systems, where simple control systems are nested within more complex ones to create adaptive behaviors​

Question 27

phylogenetic control: pragmatic representations

Answer 27

the brain represents the world in terms of actions, focusing on opportunities for interaction (affordances) rather than abstract descriptions​

Question 28

meditation: bayesian model reduction

Answer 28

• brain’s offline process of simplifying its models of the world by reducing complexity unnecessary details while still maintaining accuracy
• fact-free learning: the brain updates and simplifies its models even without new facts, allowing it to refine generative models
• trade-off between accuracy and complexity: a less complex model is more parsimonious at the expense of detail at a higher level in the hierarchy

Question 29

meditation predictions

Answer 29

1. temporal depth of mental activity decreases from FA > OM > ND
2. if experience is reduced to the present moment, this also modulates novel prediction formation and PE signalling

Question 30

prediction 1: decrease in temporal depth

Answer 30

• the temporal depth of mental activity reduces from FA to OM to ND meditation
• thereby, neural activity should decrease across the temporal hierarchy across meditation styles
• this will affect all temporally thick mental processes, including conceptualization and the sense of self

Question 31

paradigm shift

Answer 31

a fundamental change in the basic concepts and experimental practices of a scientific discipline

Question 32

4 stages of a paradigm shift

Answer 32

1. normal science
2. paradigm crisis
3. adoption of new paradigm
4. new paradigm is dominant

Question 33

current deviant observations

Answer 33

1. action signals in primary visual cortex
2. 10x as many feedback connections from V2 to V1 than feedforward connection from V1 to V2
3. blurry boundaries between cognitive concepts and between cognitive and motor regions
4. negligence of the embodied nature of the brain
5. theories separate per cognitive function, and no overarching theory

Question 34

limits of cognitive approach (why its hard to explain certain observations)

Answer 34

• stimulus-driven learning does not lead to generalization

1. lack of transfer of learning in computerized cognitive training
2. AI also learns from stimulus driven approaches and cannot transfer its knowledge

Question 35

how to induce general learning

Answer 35

1. top-down learning
2. naturalistic settings