In the last decades, the relationship between consciousness and the brain has become a fundamental topic of debate both in theoretical and empirical neurosciences. One of the most important approach for studying this relationship is based on the Integrated Information Theory (IIT) proposed by Giulio Tononi in 2004. This theory starts from the essential properties of phenomenal experience, from which it derives the requirements for the physical substrate of consciousness. According to this theory, the joint presence of functional integration and specialization at the neural network level is associated with brain complexity. On the contrary, a system that is integrated but not differentiated (or vice versa), should not be complex and should not sustain consciousness. This theoretical prediction can be empirically tested by using a perturbational approach (Marcello Massimini et al. 2009) by combining Transcranial Magnetic Stimulation (TMS) and High Density Electroencephalography (hd-EEG). TMS/hd-EEG can be used to record the immediate reaction of the brain to controlled perturbations of different cortical areas (Marcello Massimini et al. 2005), thus examining to what extent different regions of the thalamocortical system can interact causally (integration), producing specific responses (information) – i.e. a complex spatiotemporal dynamics. The Perturbational Complexity Index (PCI), proposed in 2013 by Adenauer G. Casali and colleagues, aimed at measuring the integrated information from the thalamocortical system of a subject from real data. Yet, although clinically relevant and empirically validated, these measures still lack a formal link to the theoretical framework. In this work, we propose to fill this gap by using an in-silico modelling approach based on The Virtual Brain (P. S. Leon et al. 2013) and using different a priori connectivity assumptions. Specifically, three different kinds of connectivity were taken into account by rearranging brain cortical areas’ connections, according to IIT’s postulates and possible scenarios: a system whose units were connected through the connectome where specialization and integration are balanced; an all-to-all system with low specialization and high integration and a modular system with high specialization and low integration. Cortical areas of interest were systematically activated with opportune stimulation. The aim was to investigate the brain response as a function of a long-range structural connectivity reproducing non-invasive techniques’ conditions such as TMS/hd-EEG stimulation protocol. Two versions of PCI, respectively PCIlz and the more recent PCIst, were employed as clinical classifiers to verify the measurement of information on modelled data. PCI indexes were found to be efficient in capturing the complexity of the modelled responses. PCIst and PCIlz values were higher, regardless of the stimulation site and the type of signal (i.e., stochastic or deterministic), for the connectivity characterised by a proper balance between integration and specialization, from others, in which, one of the two properties strongly characterized the network (in this case integration for the all-to-all and specialization for the modular network). Results were consistent with the theoretical predictions and they suggest the validity of the proposed approach as a first step to reconcile theory with experimental evidence. This theoretical prediction can be empirically tested by using a perturbational approach (Marcello Massimini et al. 2009) by combining Transcranial Magnetic Stimulation (TMS) and High Density Electroencephalography (hd-EEG). TMS/hd-EEG can be used to record the immediate reaction of the brain to controlled perturbations of different cortical areas (Marcello Massimini et al. 2005), thus examining to what extent different regions of the thalamocortical system can interact causally (integration), producing specific responses (information) – i.e. a complex spatiotemporal dynamics. The Perturbational Complexity Index (PCI), proposed in 2013 by Adenauer G. Casali and colleagues, aimed at measuring the integrated information from the thalamocortical system of a subject from real data. Yet, although clinically relevant and empirically validated, these measures still lack a formal link to the theoretical framework. In this work, we propose to fill this gap by using an in-silico modelling approach based on The Virtual Brain (P. S. Leon et al. 2013) and using different a priori connectivity assumptions. Specifically, three different kinds of connectivity were taken into account by rearranging brain cortical areas’ connections, according to IIT’s postulates and possible scenarios: a system whose units were connected through the connectome where specialization and integration are balanced; an all-to-all system with low specialization and high integration and an all-to-all system with low specialization and high integration. Cortical areas of interest were systematically activated with opportune stimulation. The aim was to investigate the brain response as a function of a long-range structural connectivity reproducing non-invasive techniques’ conditions such as TMS-hdEEG stimulation protocol. Two versions of PCI, respectively PCILZ and the more recent PCIST, were employed as clinical classifiers to verify the measurement of information on modelled data. PCI indexes were found to be efficient in capturing the complexity of the modelled responses. PCIST and PCILZ values were higher, regardless of the stimulation site and the type of signal (i.e., stochastic or deterministic), for the connectivity characterised by a proper balance between integration and specialization, from others, in which, one of the two properties strongly characterized the network (in this case integration for the all-to-all and specialization for the modular network). Results were consistent with the theoretical predictions and they suggest the validity of the proposed approach as a first step to reconcile theory with experimental evidence.
Negli ultimi decenni, la relazione tra coscienza e cervello è stata oggetto di discussione sia nelle neuroscienze teoriche che empiriche. Uno degli approcci più importanti, per lo studio di questa relazione, è dato dalla Teoria dell’Informazione Integrata, proposta da Giulio Tononi nel 2004. Questa teoria parte dalle proprietà essenziali dell'esperienza fenomenica, da cui derivano i requisiti per il substrato fisico della coscienza. Secondo questa teoria, la presenza congiunta di integrazione e specializzazione funzionale a livello di rete neurale è associata alla complessità del cervello. Al contrario, un sistema integrato ma non differenziato (o viceversa), non dovrebbe essere complesso. Questa previsione teorica può essere testata empiricamente usando un approccio perturbativo (Marcello Massimini et al. 2009) combinando la stimolazione magnetica transcranica (TMS) e l'elettroencefalografia ad alta densità (hd-EEG). La TMS/hd-EEG può essere usata per registrare la reazione immediata del cervello a perturbazioni controllate di diverse aree corticali (Marcello Massimini et al. 2005), esaminando in che misura diverse regioni del sistema talamocorticale possono interagire causalmente (integrazione), producendo risposte specifiche (informazioni) - cioè una complessa dinamica spazio-temporale. Il Perturbational Complexity Index (PCI), proposto nel 2013 da Adenauer G. Casali, insieme ad un gruppo di ricercatori, mirava a misurare le informazioni integrate dal sistema talamocorticale di un soggetto da dati reali. Tuttavia, sebbene clinicamente rilevanti e validati empiricamente, queste misure mancano ancora di un legame formale con il quadro teorico. In questo lavoro, la piattaforma Virtual Brain (TVB) ci viene in aiuto al fine di proporre un approccio di modellizzazione basate su ipotesi a priori a livello di connettività cerebrale. A questo scopo, è stato utilizzato un modello che considerasse l’intera struttura cerebrale, implementato da TVB, su scala macroscopica. Tre diversi tipi di connettività sono stati presi in considerazione riordinando le connessioni delle aree corticali, secondo i postulati della Teoria dell’Informazione Integrata (IIT) e i tre possibili scenari: un sistema le cui unità erano collegate attraverso un connettoma con proprietà di specializzazione e integrazione equilibrate; un sistema all-to-all con bassa specializzazione e alta integrazione e un sistema modulare con alta specializzazione e bassa integrazione. Aree corticali di interesse sono state sistematicamente attivate con opportuna stimolazione. L'obiettivo era quello di studiare la risposta cerebrale in funzione della connettività strutturale a lungo raggio riproducendo le condizioni di tecniche non invasive, quale ad esempio il protocollo di stimolazione TMS/hd-EEG. Due versioni di PCI, rispettivamente PCIlz e PCIst più recente, sono state impiegate come indicatori sperimentali clinici per verificare la complessità della risposta simulata. Gli indici PCI si sono rivelati essere efficienti nel catturare la complessità delle risposte modellate. I valori di PCIst e PCIlz sono risultati più alti, indipendentemente dal sito di stimolazione e dal tipo di segnale (stocastico o deterministico), per la connettività caratterizzata da un adeguato equilibrio tra integrazione e specializzazione, in confronto alle altre, in cui una delle due proprietà era fortemente caratterizzante per la rete (in questo caso l'integrazione per la rete all-to-all e la specializzazione per quella modulare). I risultati si sono rivelati coerenti con le previsioni teoriche, e suggeriscono la validità dell’approccio proposto come primo passo per conciliare la teoria con le evidenze sperimentali.
The role of structural connectivity for deterministic, complex spatio-temporal dynamics : a modelling approach
CALIANDRO, GIANDOMENICO
2018/2019
Abstract
In the last decades, the relationship between consciousness and the brain has become a fundamental topic of debate both in theoretical and empirical neurosciences. One of the most important approach for studying this relationship is based on the Integrated Information Theory (IIT) proposed by Giulio Tononi in 2004. This theory starts from the essential properties of phenomenal experience, from which it derives the requirements for the physical substrate of consciousness. According to this theory, the joint presence of functional integration and specialization at the neural network level is associated with brain complexity. On the contrary, a system that is integrated but not differentiated (or vice versa), should not be complex and should not sustain consciousness. This theoretical prediction can be empirically tested by using a perturbational approach (Marcello Massimini et al. 2009) by combining Transcranial Magnetic Stimulation (TMS) and High Density Electroencephalography (hd-EEG). TMS/hd-EEG can be used to record the immediate reaction of the brain to controlled perturbations of different cortical areas (Marcello Massimini et al. 2005), thus examining to what extent different regions of the thalamocortical system can interact causally (integration), producing specific responses (information) – i.e. a complex spatiotemporal dynamics. The Perturbational Complexity Index (PCI), proposed in 2013 by Adenauer G. Casali and colleagues, aimed at measuring the integrated information from the thalamocortical system of a subject from real data. Yet, although clinically relevant and empirically validated, these measures still lack a formal link to the theoretical framework. In this work, we propose to fill this gap by using an in-silico modelling approach based on The Virtual Brain (P. S. Leon et al. 2013) and using different a priori connectivity assumptions. Specifically, three different kinds of connectivity were taken into account by rearranging brain cortical areas’ connections, according to IIT’s postulates and possible scenarios: a system whose units were connected through the connectome where specialization and integration are balanced; an all-to-all system with low specialization and high integration and a modular system with high specialization and low integration. Cortical areas of interest were systematically activated with opportune stimulation. The aim was to investigate the brain response as a function of a long-range structural connectivity reproducing non-invasive techniques’ conditions such as TMS/hd-EEG stimulation protocol. Two versions of PCI, respectively PCIlz and the more recent PCIst, were employed as clinical classifiers to verify the measurement of information on modelled data. PCI indexes were found to be efficient in capturing the complexity of the modelled responses. PCIst and PCIlz values were higher, regardless of the stimulation site and the type of signal (i.e., stochastic or deterministic), for the connectivity characterised by a proper balance between integration and specialization, from others, in which, one of the two properties strongly characterized the network (in this case integration for the all-to-all and specialization for the modular network). Results were consistent with the theoretical predictions and they suggest the validity of the proposed approach as a first step to reconcile theory with experimental evidence. This theoretical prediction can be empirically tested by using a perturbational approach (Marcello Massimini et al. 2009) by combining Transcranial Magnetic Stimulation (TMS) and High Density Electroencephalography (hd-EEG). TMS/hd-EEG can be used to record the immediate reaction of the brain to controlled perturbations of different cortical areas (Marcello Massimini et al. 2005), thus examining to what extent different regions of the thalamocortical system can interact causally (integration), producing specific responses (information) – i.e. a complex spatiotemporal dynamics. The Perturbational Complexity Index (PCI), proposed in 2013 by Adenauer G. Casali and colleagues, aimed at measuring the integrated information from the thalamocortical system of a subject from real data. Yet, although clinically relevant and empirically validated, these measures still lack a formal link to the theoretical framework. In this work, we propose to fill this gap by using an in-silico modelling approach based on The Virtual Brain (P. S. Leon et al. 2013) and using different a priori connectivity assumptions. Specifically, three different kinds of connectivity were taken into account by rearranging brain cortical areas’ connections, according to IIT’s postulates and possible scenarios: a system whose units were connected through the connectome where specialization and integration are balanced; an all-to-all system with low specialization and high integration and an all-to-all system with low specialization and high integration. Cortical areas of interest were systematically activated with opportune stimulation. The aim was to investigate the brain response as a function of a long-range structural connectivity reproducing non-invasive techniques’ conditions such as TMS-hdEEG stimulation protocol. Two versions of PCI, respectively PCILZ and the more recent PCIST, were employed as clinical classifiers to verify the measurement of information on modelled data. PCI indexes were found to be efficient in capturing the complexity of the modelled responses. PCIST and PCILZ values were higher, regardless of the stimulation site and the type of signal (i.e., stochastic or deterministic), for the connectivity characterised by a proper balance between integration and specialization, from others, in which, one of the two properties strongly characterized the network (in this case integration for the all-to-all and specialization for the modular network). Results were consistent with the theoretical predictions and they suggest the validity of the proposed approach as a first step to reconcile theory with experimental evidence.| File | Dimensione | Formato | |
|---|---|---|---|
|
12_2019_Caliandro.pdf
solo utenti autorizzati dal 03/12/2022
Descrizione: Testo della tesi
Dimensione
4.26 MB
Formato
Adobe PDF
|
4.26 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/151098