Propofol induction reduces the capacity for neural information integration: Implications for the mechanism of consciousness and general anesthesia

https://doi.org/10.1016/j.concog.2008.10.005Get rights and content

Abstract

The cognitive unbinding paradigm suggests that the synthesis of neural information is attenuated by general anesthesia. Here, we analyzed the functional organization of brain activities in the conscious and anesthetized states, based on functional segregation and integration. Electroencephalography (EEG) recordings were obtained from 14 subjects undergoing induction of general anesthesia with propofol. We quantified changes in mean information integration capacity in each band of the EEG. After induction with propofol, mean information integration capacity was reduced most prominently in the γ band of the EEG (p = .0001). Furthermore, we demonstrate that loss of consciousness is reflected by the breakdown of the spatiotemporal organization of γ waves. We conclude that induction of general anesthesia with propofol reduces the capacity for information integration in the brain. These data directly support the information integration theory of consciousness and the cognitive unbinding paradigm of general anesthesia.

Introduction

Numerous theories and recent data suggest that the synthesis of cognitive information is an essential requirement for consciousness (Crick and Koch, 1995, Hameroff, 1996, Mashour, 2004, Mashour, 2006, Singer, 1996). Integrative neural activities such as recurrent processing from higher to lower cortical areas, as well as coherent 40 Hz oscillations, have been associated with higher cognitive function or consciousness (Mashour, 2006). There has been a particular focus on functional and effective connectivity of the thalamocortical and corticocortical systems due to their ability to integrate the activities of functionally diverse cognitive modules (Tononi and Sporns, 2003, Tononi, 2004). Accordingly, Tononi’s “information integration theory” formalizes this synthetic property as the neural basis of consciousness. Tononi has created a mathematical model of the capacity for information integration, denoted as Φ, which is increased in systems maintaining consciousness (e.g., thalamocortical processes) and decreased in those that do not (e.g., cerebellar processes, sleep).

Since the integration or binding of neural processes is deemed essential to the generation of conscious states, it has been suggested that the “unbinding” of these processes may be essential to the generation of unconscious states (John and Prichep, 2005, Mashour, 2004, Mashour, 2006). The cognitive unbinding paradigm of general anesthesia suggests that cognitive activity does not need to be eliminated, but simply disintegrated, for general anesthetics to interrupt consciousness. There has been indirect evidence and support of this theory. Anesthetic-mediated loss of consciousness has been associated with decoherence of activity around 40 Hz using quantitative EEG (John et al., 2001), loss of thalamocortical connectivity using positron emission tomography (PET) (Alkire, Haier, & Fallon, 2000), and loss of functional connectivity using magnetic resonance imaging (MRI) (Peltier et al., 2005). These findings are consistent with the loss of effective cortical connectivity associated with sleep (Massimini et al., 2005). What has not yet been reported is a direct measure of the capacity for neural information integration in the conscious and anesthetized state. A decrease in Φ associated with the induction of general anesthesia would be a direct demonstration of the importance of information integration in conscious processes and would support cognitive unbinding as one mechanism of anesthetic-mediated unconsciousness. The task is to translate these theoretical paradigms into empirical studies in humans. In order to accomplish this, we developed a novel method of human EEG reconstructions in state space.

State space (or phase space) is the space in which all possible states of a dynamic system (such as the anesthetized brain) can be represented. In state space, each parameter of a system is represented as an axis of a multi-dimensional space. Consider a classic example from physics: the swinging pendulum. The location of the pendulum can be plotted in structural space, but its overall behavior as a system is plotted in state space. The back-and-forth motion in “real space” is what we would actually see if a pendulum were swinging. The state space of the swinging pendulum, however, would show the oscillating behavior of the system using the parameters of position and velocity, unfolding over the parameter of time. This behavior would translate to a sine wave plotted in three dimensions, which represent the three parameters. When a system converges to a typical behavior, the corresponding sets in state space are referred to as the attracting sets or attractor. Attractors may be linear or non-linear: when they are described by fractal rather than Euclidean geometry, they are said to be strange attractors.

Functional integration (“binding”) and segregation in the brain are fundamental principles of neural organization (Hameroff, 1996, Mashour, 2004, Tononi and Sporns, 2003, Tononi, 2004). Here we assumed that the functional integration and segregation of the brain is different in conscious and anesthetized states. In order to investigate this functional organization in conscious and unconscious states, we developed a new method based on the information integration theory and state space reconstruction method. By using the state space reconstruction method, a complex combination of several functional segregations and integrations in the brain can be represented as a point in a state space. Furthermore, the temporal evolution of such combinations can be reconstructed as a trajectory, much like the sine wave of the pendulum’s behavior moves forward in time. With this reconstructed trajectory in a state space, we can consider cognitive states as a dynamic problem of spatiotemporal functional organization of the brain. This technique also enables quantification of the functional characteristics in conscious and anesthetized brain with several measures in nonlinear science. For example, the dimension of a reconstructed state space corresponds to the degree of freedom of a system.

There is a precedent for using state space and EEG to study consciousness and anesthesia. Watt and Hameroff (1988) demonstrated that state space analysis of EEG reveals distinct attractors and dimensions for distinct states of consciousness, while van den Broek et al. (2006) showed that dimensionality and complexity are measures of anesthetic depth. Finally, Walling and Hicks (2006) found that emergence from sevoflurane was associated with a transition from an ordered attractor of unconsciousness to a chaotic strange attractor of wakefulness. Of interest, they deemed emergence from anesthesia as a “cognitive rebinding.” These previous studies considered the dynamic properties of single EEG channel or the average of single EEG channels’ properties. In contrast, we take into account the spatiotemporal organization of segregated functions. Each segregated function was determined by a computational method, called “minimum information bipartition”, as a cluster of EEG channels.

Here, we report that the information integration capacity (Φ) of subjects is significantly reduced after induction with propofol. We also show the spatiotemporal organization of segregated functions is distinct in states of consciousness and general anesthesia. This distinctive feature appears only in the γ band, which has been associated with cognitive information processing in the brain.

Section snippets

Subjects

After obtaining the approval of the Institutional Review Board of Asan Medical Center (Seoul, Korea) and written informed consent, we studied 14 patients, aged 20–80 years, who were scheduled for stomach surgery. Given the higher incidence of gastric cancer in this region and associated early screening, patients were identified prior to any signs or symptoms of mechanical obstruction and were therefore scheduled for elective cases with routine induction of anesthesia. All patients were American

State space reconstructions

Fig. 1(a) and (b) demonstrate state space construction from multi-channel EEG. Fig. 1(a) depicts three representative EEG complexes, determined by the MIB method, while Fig. 1(b) depicts the point reconstructed by the three EEG complexes. Fig. 1(c) shows the two trajectories, which represent different dynamics of different EEG complexes. Fig. 1(d) shows the sequences of the three EEG complexes selected from a subject. Fig. 1(e) and (f) demonstrate a state space and the decomposed space by the

Discussion

Here, we report that the general anesthetic propofol is associated with a decrease in the capacity for information integration in the human brain. Our data further suggest that states of human consciousness and unconsciousness are directly associated with the information integration capacity of γ waves. These data agree well with the predictions of the information integration theory of consciousness (Tononi and Sporns, 2003, Tononi, 2004), as well as the cognitive unbinding paradigm of general

Conflicts of Interest

The authors have no conflicts to declare.

Acknowledgments

This study was supported by a grant of the Korea Health 21 R&D Project (A060775), Ministry of Health & Welfare, and a grant of the Brain Korea 21 Project, Ministry of Education & Human Resources, Republic of Korea.

References (28)

  • S. Hameroff

    Conscious events as orchestrated space-time selections

    Journal of Consciousness Studies

    (1996)
  • E.R. John et al.

    The anesthetic cascade: A theory of how anesthesia suppresses consciousness

    Anesthesiology

    (2005)
  • D. Kugiumtzis

    Surrogate data test for nonlinearity including nonmonotonic transforms

    Physical Review E

    (2000)
  • G.A. Mashour

    Consciousness unbound: Toward a paradigm of general anesthesia

    Anesthesiology

    (2004)
  • Cited by (111)

    • The strength of weak integrated information theory

      2022, Trends in Cognitive Sciences
      Citation Excerpt :

      There is evidence for some Φ measures decreasing during loss of consciousness, (e.g., Φ⁎ in local field potentials [48], Φ˜ in EEG [49], and ΦR in fMRI [11]). However, there are also some conflicting findings, for example, Lee et al. [50] report decreased ΦWMS under anaesthesia in the gamma band, while Kim et al. [51] report lower Φ⁎ in the alpha band but higher Φ⁎ in all other frequency bands. To make sense of these mixed results, computational work has shown that even in simple systems these measures can behave very differently [45], so it is not surprising they differ when applied to real data too.

    View all citing articles on Scopus
    View full text