Bassett, D.S. 2017¶
§1 Learning as a Network Phenomenone¶
¶1
¶2
a new literature is emerging that focuses on the effects of learning at a coarser level – between entire brain regions
¶3
broad-scale changes in neurophysi-ological dynamics across distributed neural circuits or networks
When learning produces such distributed network changes–refl ected either in anatomy or function– it is useful to consider quantitative methods that can not only describe that network architecture but also predict its dynamics
¶4
In this review...
- unique approach to describing neural systems in terms of associations in the brain
- the reconfigurations underlying its adaptive processe
- how network neuroscience could provide a quantitative framework that complements existing models of learning by cohesively accounting for network structure in neurophysiological and behavioral data
§2 Network Neuroscience¶
¶1
Network science: a subfield of complex systems science
¶2
"network neuroscinece" emcompass:
- the use of network science in understanding connectivity patterns
- the impact of these connectivity patterns on animal behavior
describe the architecture of relational data
¶3
E,
V,
forming a dyad,
dyads can change in their strength over time
¶4
- mesurement of local structure
- measurement of mesoscale & global structure
local
: e.g., clustering coeeficient of graph
mesoscale
: e.g., modularity of G (measures the presence/strength of local clusters of interconnected nodes)
¶5
G structure can have distincet implications for how the system functions
For e.g,
Random G
: N has equal prob of connecting to any other Ns
: transmit info quickly, NOT local info integration
Regular G
: N connetcts to equal # of neighbor
: local info integration, NOT global transmission of info
Modular G
: evolvabiloty
: contain groups of Ns that can change / adapt their func w/o perturbin other groups
Hierachial modular network
: similat to human brain
: specialization of nested func & adaptability
¶6
functional brain network
: fMRI, MEG, EEG
: Ns - region/voxel; Es - functional connectivity
infer circuits
¶7
DWI
Structural network
: WM microstructure
: Vs - ; Es - estimated strength of WM tracts
§3 Dynamic Networks During Learning¶
¶8
Temporal network
: each G represents interaction pattern in single time-window
: ensemble of Gs - evlution of interaction patterns
N in one time-window is connected to iteself in neighboring time-windows using interlayer link
- traditional stat : adjacency matrix
- new stat : adjacency tensor
¶9
- Within-scan : functional imaging
- Across-scan : functiona & structural imaging
¶10
Base on Hebbian learning,
Δ in struc & func: driven by patterns of activity and connectivity
functional connectivity e.g., 1. detect memory rehabilitation trainning intervention following stroke 2. predict future abilty of individual to learn motor skills 3. words of artificial language
¶11
motor skill learning
network reconfigulation (learning) * struc Δ: * func Δ :
§4 Reconfigulation of Network Modulates During Learning¶
¶12
in motor skill learning, network stat Δ:
↑ clustering coefficients
↑ # of network connections
↑ connection strength
↓ communication distances
* Δ in network centrality
learning requires Δ in modular organization
Modular network
: contains local clusters of densely interconnected Ns
recruitments of & integration between systems is altered
¶13
reconfigulation are charcteristics of a flexible brain networks organizaition individual Δ predicts future learning rate
¶ 14
modularity plays role in behavioral adaptability
¶15
modularity is marker of high congnitive func
- modularity ∝ working memory
- flexible rearrangement ∝ Δ memory acuracy & cognitive flexibility
§5 Challenges & Oppotunities¶
¶16
? States / Traits
? depends on the type of learning
? Can predict
? brains flexible
? modulate w/ mood / pharmochological intervention
? modulated by NMDA / NE / other
¶17