Attention: A Complex System
From the Intricate Modulation of Tuned Responses Towards a Layered Cortical Circuit Model
by Markus Helmer
Date of Examination:2015-09-11
Date of issue:2015-12-21
Advisor:Prof. Dr. Theo Geisel
Referee:Prof. Dr. Theo Geisel
Referee:Prof. Dr. Fred Wolf
Persistent Address:
http://dx.doi.org/10.53846/goediss-5441
Files in this item
Name:thesis_publishedVersion_fastWebView_17dec2015.pdf
Size:11.4Mb
Format:PDF
Description:Markus Helmer PhD thesis: analysis of complex tuning curves & model for a canonical local cortical microcircuit
The following license files are associated with this item:
Abstract
English
Experiments have elucidated the neuronal correlates of attention in the primate brain, and have led to a number of models capturing various aspects of these experiments. However, these model propose qualitatively different ways of how attention acts, and often stay unclear as to how attention emerges from the biological constituents. In the first part of this work we have further demonstrated novel attentional modulation patterns, that are highly cell- and stimulus-specific and lead to-over the population-complex, non-multiplicative shape changes of tuning curves, that don't seem to be compatible with any current phenomenological (like the biased competition or the feature-similarity-gain model) or circuit model (like the Ardid-Wang-Compte model) of attention. Whereas phenomenological models fit experimental observations into an abstract, high-level description, circuit models aim at describing these data as an emergent property of the interaction of suitably chosen low-level constituents. These interactions can be constrained in a principled way through better and better fine-grained connectivity data and based on this structural skeleton complex dynamics might emerge. For example, oscillations and their interdependence have been hypothesized to play a role in the coordination between brain network constituents and, moreover, given a fixed structural skeleton, the circuit might possess a multitude of states, due to multistability or, more profanely, due to a variation in parameters like coupling efficacy or neuromodulators. We hypothesize, thus, that if the circuit model is sufficiently good, it will possess states with dynamical fingerprints resembling functional neuronal correlates, like those occuring during attention. While we are far from such a brain-wide circuit model for attention, we have investigated in the first part of this work a multi-ring circuit model to reproduce the attentional population effects mentioned above, without yet, however, achieving satisfactory results. Moreover, given the prominent role that cortical rhythms are hypothesized to play in interareal coordination, in the second part of this work, we investigate oscillations in a simple rate model for a cortical column with realistic interlayer connectivity, observing complex layer-specific multi-frequency oscillations, with upper and lower layers oscillating predominantely at fast (gamma-like) and slow (beta-like) frequencies, in line with experimental findings and suggesting that the cortical column might form an important building block for communication-through-coherence processes which are modulated by attention. We show further, that this pattern of oscillations depends crucially on this, or some structurally degenerate, connectomes, arguing against the arbitrariness of structure for brain function. Moreover, when two columns at different hierarchical levels are coupled, we obtained preliminary results indicating that a self-organized directed coupling can emerge that is ``feedforward'' in the gamma- but ``feedback'' in the beta-band, in line with a currently discussed role of cortical oscillations. In the future, building on a more systematic understanding of the two-columns system, we aim to study emerging multi-frequency oscillations in a brain-wide model with realistic, layer-resolved topology and their potential for interareal coordination and information processing in order to eventually obtain a better understanding of how the capacity to pay attention, and its neuronal correlates, emerges in the brain.
Keywords: attention; data analysis; model; cortical layers; canonical microcircuit; cortical oscillations; tuning curves; model-free feature extraction; ring model; homeostasis; contextual modulation; inter-areal coordination; communication-through-coherence