Using information theory to measure the emergence of artificial free will in a spiking brain-constrained model of the human cortex
Tools
Tools
Bourne, Josh; Rosas, Fernando, E. and Garagnani, M.. 2023. 'Using information theory to measure the emergence of artificial free will in a spiking brain-constrained model of the human cortex'. In: 32nd Annual Computational Neuroscience Meeting. Leipzig, Germany 15 - 19 July 2023. [Conference or Workshop Item]
|
Text
Abstract.pdf - Supplemental Material Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (151kB) | Preview |
Abstract or Description
Cell Assembly (CA) circuits are known to emerge in neurocomputational models as a result of Hebbian-like learning. Intriguingly, when a brain-like architecture is used, CAs spontaneously "ignite" in absence of any stimulus, and the patterns of network activation occurring during such ignitions closely match those observed in the human brain during non-stimulus driven, endogenous decisions to act [1]. This suggests that sub-threshold reverberation of noise within CA circuits (which drives their ignition) may be a possible mechanism underlying seemingly "free" and volitional (yet possibly pre-consciously determined) action decisions [2]. It is unclear, however, whether such spontaneous CA ignitions are truly an emergent property of the brain-like model, or whether they are somehow "pre-encoded" in the system's features. Can we provide objective evidence supporting (or falsifying) the hypothesis that these spontaneous events are de facto non pre-determined and can be thus be considered as the network's own endogenous "action decisions"?
To investigate this issue, we used a spiking brain-constrained model of six cortical areas and, after replicating the previously documented CA emergence and spontaneous ignitions in it, we analysed its emergent properties using information theoretic measures. Recent techniques in information theory allow quantifying emergence in complex systems (including the brain) [3]. Here, we applied these measures to test for the presence of emergence during spontaneous, unprovoked CA circuit ignition. Specifically, we analysed the different modalities of emergence associated with cell assembly ignition and lifecycle (downward causation and causal decoupling).
Preliminary results show the highest levels of emergent behaviour (specifically, causal decoupling) during cell assembly ignition, which gradually fade as CA activation dissipates. Such increased levels of causal decoupling observed during (and prior to) CA ignition episodes confirm the presence of an emergent feature in the neural model.
In summary, we present here the application of formal criteria used for determining the presence of emergence in complex systems to a spiking, brain-constrained neurocomputational model of the cortex that can mechanistically explain the neural origins of so-called "free", volitional action decisions. Initial results of the information-theoretical analysis indicate that spontaneous CA circuit ignitions, driven by reverberation of noise within them, truly constitute an emergent feature of the brain-like architecture, suggesting that this phenomenon should be considered as an endogenous (i.e., internally generated, and not pre-determined) feature of the artificial neural network.
References:
[1] Garagnani, M. & Pulvermüller, F. (2013) Neuronal correlates of decisions to speak and act: spontaneous emergence and dynamic topographies in a computational model of frontal and temporal areas. Brain and Language 127(1):75–85.
[2] Schurger A, Mylopoulos M, Rosenthal D. (2016) Neural antecedents of spontaneous voluntary movement: a new perspective. Trends in Cognitive Sciences. 20(2), 77-79.
[3] Rosas F.E., Mediano P.A.M., Jensen H.J., Seth A.K., Barrett A.B., Carhart-Harris R.L., et al. (2020) Reconciling emergences: An informationtheoretic approach to identify causal emergence in multivariate data. PLoS Comput Biol 16(12): e1008289.
|
Item Type: |
Conference or Workshop Item (Poster) |
||||||
| Related URLs: |
|
||||||
| Departments, Centres and Research Units: |
|||||||
| Dates: |
|
||||||
| Event Location: |
Leipzig, Germany |
||||||
| Date range: |
15 - 19 July 2023 |
||||||
| Item ID: |
33691 |
||||||
| Date Deposited: |
26 Jun 2023 08:36 |
||||||
| Last Modified: |
17 Jul 2023 22:26 |
||||||
|
URI: |
View statistics for this item...
 |
Edit Record (login required) |