7 
Spatiotemporal Model of Consciousness I: Spatiotemporal Specificity and Neuronal-Phenomenal Correspondence

Introduction

General Background

Consciousness is a complex phenomenon that includes different dimensions. The initial characterization of consciousness by contents (Crick & Koch, 2003; Koch, 2004) has been complemented by the level or state of consciousness (Bachmann & Hudetz, 2014; Koch et al., 2016; Laureys, 2005). Recently, additional dimensions have been suggested. One such dimension is the distinction between phenomenal/experiential and cognitive aspects of consciousness (Cerullo, Metzinger, & Mangun, 2015; Northoff, 2014b). Another dimension was introduced with the form (or structure) of consciousness (Northoff, 2013, 2014b). The form of consciousness pertains to the grouping and ultimately the organization of different contents, which, neuronally, is supposedly associated with the spontaneous activity and its spatiotemporal structure. The exact neuronal mechanisms underlying the different dimensions of consciousness, for example, level/state, content/form, phenomenal/experiential, cognitive/reporting, including their relations, remain an open question.

Many studies in healthy subjects sought to associate consciousness with stimulus-induced or task-evoked brain activity. Specifically, the stimulus-induced or task-evoked activities refer to those neural activity changes that are related to and sufficient for the contents of consciousness (Koch et al., 2016). We therefore speak of content–neural correlates of consciousness (NCC). Temporally, the content–NCC is associated with event-related potentials such as the N100, P300 (Bachmann & Hudetz, 2014; Dehaene & Changeux, 2011; Koch et al., 2016). Spatially, stimulus-induced or task-evoked activity in higher-order brain regions such as the prefrontal cortex and posterior cortical “hot zones” may be the NCC for mediating conscious content (Dehaene et al., 2014; Dehaene & Changeux, 2011; Koch et al., 2016).

More recently, different components of stimulus-induced activity have been identified including the distinction between early and late stimulus-induced activity, as well as the interaction between pre- and poststimulus activity. Early stimulus-induced activity, as tested for in “no-report” paradigms may be related to the phenomenal features of consciousness (e.g., experience), whereas late stimulus-induced activity is supposedly more related to its cognitive components (e.g., reporting and awareness of contents) (Koch et al., 2016; Lamme, 2010a,b; Northoff, 2014b; Tononi et al., 2016; Tsuchiya et al., 2015). On the other end, prior to stimulus onset, several studies have demonstrated that the level of prestimulus spontaneous activity impacts both stimulus-induced activity and the respectively associated content of consciousness (Boly et al., 2007; Hesselmann, Kell, & Kleinschmidt, 2008; Mathewson et al., 2009; Ploner et al., 2010; Qin et al., 2016; Sadaghiani, Hesselmann, Friston, & Kleinschmidt, 2010, Sadaghiani et al., 2015; Schölvinck et al., 2012; van Dijk et al., 2008; Yu et al., 2015). The relevance of prestimulus activity level suggests a central role of the brain’s spontaneous activity for consciousness. This is also supported by other studies in subjects with altered states of consciousness, such as unresponsive wakefulness state (UWRS), sleep, and anesthesia; these subjects showed major changes in the brain’s spontaneous activity (Bayne, Hohwy, & Owen, 2016).

Why and how are these different forms of neural activity (i.e., spontaneous, prestimulus, early, and late stimulus-induced activity) related to consciousness and its different dimensions? To date, this subject has not yet been thoroughly examined. I here propose that these different forms of neural activity reflect different ways of how the brain constructs its own inner time and space, that is, its intrinsic time and space (see the section below titled “Definition of Time and Space”). This hypothesis comprises what I describe as the spatiotemporal theory of consciousness (STC).

Time and Space

Time and space are the central and most basic building blocks of nature. Time and space can be constructed in different ways. Although the different ways of constructing time and space have been investigated extensively in the field of physics, their relevance for the brain’s neural activity and, even more importantly, to consciousness remains largely unknown. Current neuroscientific views focus mainly on information, behavioral, affective, or cognitive features of brain and consciousness such as integrated information theory (IIT) (Tononi et al., 2016), or cognitive theory such as the global neuronal workspace theory (GNWT) (Dehaene et al., 2014; Dehaene & Changeux, 2011), and predictive coding (see chapters 3 and 6 of this volume for details, as well as Friston, 2010; Hohwy, 2013; Seth, 2014). Whereas these views presuppose and implicitly touch on the brain’s own time and space, they do not consider time and space themselves—central dimensions of the brain’s neural activity—in an explicit way, that is, they do not consider how the brain itself constructs time and space in its own neural activity.

Given that (1) time and space are most basic features of nature and (2) that the brain itself is part of nature, we here consider the brain and its neural activity in explicitly temporal and spatial terms. In other words, we conceive the brain’s different forms of neural activity (spontaneous, prestimulus, early and late stimulus-induced activity) in primarily spatiotemporal terms rather than in informational, behavioral, cognitive, or affective terms. I postulate that such a spatiotemporal view of the brain’s neural activity is central for understanding how the brain can generate consciousness with its different dimensions. In this sense, then, consciousness may be understood as a spatiotemporal phenomenon of the brain’s neural activity.

Aim and Overview

The main and overarching aim of this chapter is to provide a unified hypothesis that directly links and thus integrates the different forms of neural activity with the different dimensions of consciousness. Such an integrative, coherent framework is suggested to consist of temporal and spatial features of the brain’s neural activity (across all forms of neural activity) comprising what I describe as the spatiotemporal theory of consciousness (STC; see also Northoff & Huang, 2017). Based on various lines of empirical evidence, I postulate here that the four dimensions of consciousness (level/state, content/form, phenomenal/experience, cognitive/reporting) are mediated by four corresponding spatiotemporal neuronal mechanisms: (1) the neuronal mechanism of spatiotemporal nestedness accounts for the level or state of consciousness; (2) the neuronal mechanism of spatiotemporal alignment accounts for selecting the content and constituting the structure, or, as I say, the form of consciousness (Northoff, 2013, 2014b); (3) the neuronal mechanism of spatiotemporal expansion accounts for the phenomenal dimension of consciousness, for example, experience with qualia; (4) the neuronal mechanism of spatiotemporal globalization accounts for the cognitive dimension of consciousness, that is, the reporting of its contents (see a summary in figure 7.1). The STC is primarily a neuroscientific theory of brain and consciousness, which carries major philosophical implications of a novel view of consciousness (see also Northoff, 2014b, 2016a–d, 2017a,b). Most importantly, the spatiotemporal view of consciousness entails a paradigm shift from a mind–body problem to a world–brain problem, which will be explicated in chapters 9–11 of this book (figure 7.1).

Three of these mechanisms (1, 3, and 4) I discuss in detail in the present chapter and reserve a detailed analysis of the fourth, spatiotemporal alignment, for the next chapter (chapter 8) in the volume. Conceptually, I also discuss two arguments in the third part of the present chapter that may be raised against such a spatiotemporal model of consciousness. First, the argument of nonspecificity claims that the suggested spatiotemporal mechanisms remain unspecific with respect to consciousness. I will reject that argument by showing that consciousness is based on specific spatiotemporal mechanisms of the brain’s neural activity. I therefore characterize consciousness by spatiotemporal specificity, as will be discussed in the first part in this chapter.

Second, the argument of triviality states that the characterization of both brain and consciousness in spatiotemporal terms remains trivial because time and space are the basic ingredients of the world in general and of the brain in particular. I will reject that argument by demonstrating a nontrivial correspondence between neuronal and phenomenal features, that is, a neuronal-phenomenal correspondence, with respect to spatiotemporal features, as will be discussed in the second part of this chapter.

Definition of Time and Space

What do we mean by the terms time and space? One would argue that the brain’s neural activity is by default temporal and spatial. This makes any account of consciousness and its different dimensions in spatiotemporal terms rather self-evident. We do need, however, to clarify what exactly we mean by the concepts of time and space in STC (see also chapter 9 for an ontological definition of time and space).

The STC refers to time and space of the brain; that is, how the brain constructs its own time and space in its neural activity. One may thus speak of time and space of the brain, or the “intrinsic” time and space of its neural activity. The construction of such intrinsic time and space of the brain itself and its neural activity needs to be distinguished from our perception and cognition of time and space including their neural correlates—the latter presupposes the former. The focus in this chapter is not on the neural correlates of our perception and cognition of time and space but rather on how the brain itself constitutes its own time and space, that is, intrinsic time and space (see appendix 2 in Northoff, 2014b, for more details).

The brain’s intrinsic time concerns the duration of neuronal activity embedded in specific frequency ranges. These frequency ranges are distinguished from higher frequency ranges such as ultrasound (Nagel, 1974) or lower frequency ranges of other nature phenomena such as seismic earth waves (He et al., 2010). As for the brain’s intrinsic space, we speak of the extension of neural activity across different regions and networks in the brain. Briefly, the brain’s intrinsic time and space or its “operational time and space” (Fingelkurts, Fingelkurts, & Neves, 2013), can be characterized by temporal duration and spatial extension of its neural activity. Both terms, temporal duration and spatial extension describe how the brain constitutes its self, that is, by “intrinsic” neural activity within the space and time of the anatomostructural brain.

Let us describe the notions of temporal duration and spatial extension of the brain’s neural activity in more empirical detail. First, the temporal duration is related to the temporal ranges or circle durations of neural oscillations or fluctuations. This includes different frequencies ranging from infraslow (0.0001–0.1 Hz), over slow (0.1–1 Hz), delta (1–4 Hz), and theta (5–8 Hz) to faster frequencies of alpha (8–12 Hz), beta (13–30 Hz), and broadband gamma (30–240 Hz) (Buzsáki, 2006; Buzsáki et al., 2013; Buzsáki & Draguhn, 2004). These different frequencies show different functions and, most likely, are associated with different underlying neurophysiological mechanisms that give rise to a wide range of behavioral and functional opportunities (Buzsáki, 2006).

Second, the temporal duration of the brain’s neural activity can also be characterized by intrinsic temporal autocorrelation in milliseconds to seconds and minutes range. These time scales can be measured by an “auto-correlation window” (Honey et al., 2012) and scale-free or fractal properties such as the power law exponent or Hurst exponent (He, 2014; He et al., 2010). This characterization of the brain’s neural activity by temporal duration across different time scales makes clear that the brain’s intrinsic time (i.e., its inner duration) is highly structured and very finely organized. We will see further below that such temporal structure and organization in the brain’s neural activity are central for consciousness.

Third, the range of frequencies and the intrinsic temporal organization of the brain’s neural activity strongly influence the processing of extrinsic stimuli. The different frequencies with their respective cycle durations provide windows of opportunity, that is, “temporal receptive windows” (Hasson, Chen, & Honey, 2015) to acquire and encode extrinsic stimuli and their temporal sequences (Lakatos et al., 2008, 2013; Schroeder & Lakatos, 2009a,b). Therefore, there appear to exist intrinsic “temporal receptive windows” that match with the physical features of the extrinsic stimuli in hierarchy time scales (Chen, Hasson, & Honey, 2015; Hasson et al., 2015; Honey et al., 2012; Murray et al., 2014).

We may now want to consider the intrinsic space of the brain’s neural activity, that is, its spatial extension. The brain shows an extensive structural connectivity that links and connects across neurons, regions, and networks. This structural connectivity provides the hardware through which neurons can functionally communicate (called functional connectivity). Although there is strong dependency of functional connectivity on structural connectivity (Honey et al., 2009), the divergence between both forms of connectivity is relevant for consciousness: loss of consciousness is marked by a loss of divergence between structural and functional connectivity (Tagliazucchi et al., 2016). Finally, it should be mentioned that the brain’s intrinsic space is also related to its small-world organization (one that is spatially scale-free) with various features including modularity and centrality (Bassett & Sporns, 2017; Sporns & Betzel, 2016).

We have so far described the temporal duration and spatial extension of the brain’s neural activity and how together they construct the brain’s intrinsic time and space. We also need to consider that the brain and its intrinsic time and space are located in the extrinsic time and space encompassing both the body and the world (Park, Correia, Ducorps, & Tallon-Baudry, 2014; Park & Tallon-Baudry, 2014). Empirical data suggest that the brain’s intrinsic time and space align themselves to extrinsic time and space in order to constitute a world–brain relation (see the next section for details of such spatiotemporal alignment). Such world-brain relation allows us to experience ourselves including our body within and our self as part of the spatiotemporally more extended world.

I propose that the brain constitutes the temporal and spatial features of its own neural activity in a most specific way—I therefore speak of what I describe as temporal and spatial mechanisms. Most importantly, I propose that these spatiotemporal mechanisms with their construction of the brain’s duration and extension are central for constituting the different dimensions of consciousness, the level/state, content, and form. In a nutshell, the spatiotemporal model of consciousness conceives both brain and consciousness in spatiotemporal terms—I propose that a specific way of constituting time and space by the brain’s neural activity is central for transforming neural activity into phenomenal activity, what we call consciousness.

Spatiotemporal Model of Consciousness I: Spatiotemporal Nestedness—Neural Predisposition of Consciousness and Spatiotemporal Specificity

Empirical Findings Ia: Spontaneous Activity—Temporal Nestedness

The brain’s spontaneous activity shows a certain temporal structure. This is, for example, reflected in the fact that the spontaneous activity operates across different frequency ranges (as from infraslow/0.01–0.1 Hz) over slow (0.01–1 Hz) and fast (1–180 Hz) ranges (Buzsáki, 2006; Buzsáki et al., 2013). Importantly, neural activity in the different frequencies shows a fractal organization such that the power of slower frequency ranges is higher than that in faster frequency ranges, which can be described as scale-free activity (He, 2014; He et al., 2010; Linkenkaer-Hansen et al., 2001; Palva et al., 2013; Palva & Palva, 2012; Zhigalov et al., 2015).

Scale-free activity has been further characterized by long-range temporal autocorrelation (LRTC) across widespread cortical regions (Bullmore et al., 2001; He 2011; He et al., 2010; Linkenkaer-Hansen, Nikouline, Palva, & Ilmoniemi, 2001; Palva et al., 2013). As fluctuations, the infraslow frequencies have often been conceived as mere noise; however, that 1/f noise-like signal consists of the neural activity itself (rather than to some noise-related activity by our method of measurement). The data as shown below suggest that the structured 1/f noise-like signal is central for the level/state of consciousness.

In addition to their scale-free fractal nature with LRTC, infraslow, slow, and faster frequencies are also coupled to each other as measured by cross-frequency coupling (CFC) (Aru et al., 2015; Bonnefond et al., 2017; He, Zempel, Snyder, & Raichle, 2010; Hyafil et al., 2015). CFC allows for linking the different frequency ranges, typically with the amplitude of the higher-frequency ranges being coupled to and integrated within the phase of the lower-frequency ones. This has been shown in both slow/faster (Aru et al., 2015; Buzsáki & Draguhn, 2004; Buzsáki et al., 2013; Hyafil, Giraud, Fontolan, & Gutkin, 2015) and infraslow ranges (Huang et al., 2016).

To summarize, the brain’s spontaneous activity shows an elaborate temporal organization in that different temporal ranges or scales are linked and integrated with each other. This is manifest in scale-free activity with LRTC and fractal nature as well as in CFC. Such organization amounts to what can be described as temporal nestedness of the brain’s spontaneous activity.

Empirical Findings Ib: Spontaneous Activity—Spatial Nestedness

We may now consider the spatial organization of the brain’s spontaneous activity. The faster frequencies are relatively spatially restricted in the brain and temporally regular (Buszaki, 2006). In contrast, the infraslow frequencies are more spatially extended throughout the brain and temporally irregular (Buzsáki, 2006; He, 2014; He et al., 2010). Moreover, spatially, one can observe modularity with small-world properties, which also follows scale-free fractal organizational patter on the spatial level (Sporns & Betzel, 2016)—this allows for hierarchical modularity with both integration and segregation of information across the whole brain (Deco et al., 2015).

Moreover, different regions or networks show different time scales. For instance, the sensory regions/networks show rather short time scales, whereas the default-mode network (DMN) seems to exhibit the longest time scales and thus the strongest power in the infraslow frequency range (see Hasson et al., 2015; Lee et al., 2014; He 2011; Huang, Zhang, Longtin, et al. 2017). Therefore, analogous to the temporal side, one may want to speak of spatial nestedness with a hierarchy of time scales (Murray et al., 2014).

Taken together, scale-free activity with LRTC, CFC (as well as other temporal features including variability, complexity, or cross-threshold crossing), and small-worldness with hierarchical modularity establish a specific temporal and spatial structure in the brain’s overall or global neural activity (Huang et al., 2014, 2016; Hudetz et al., 2015, Mitra et al., 2015; Palva et al., 2013; Palva & Palva, 2012; see also He et al., 2010).

Due to the nesting of different frequencies/regions in such spatiotemporal structure, we characterize the latter by the term spatiotemporal nestedness. Such spatiotemporal nestedness may be described as an “integrated hierarchy of time and spatial scales” (Murray et al., 2014; see also Bonnefond et al., 2017; Florin & Baillet, 2015). We see in the following section that such an integrated hierarchy of time and spatial scales of the brain’s neural activity is central for the level/state of consciousness.

Empirical Findings IIa: Spatiotemporal Nestedness and the Level/State of Consciousness

We must also consider the LRTC, CFC, and small-worldness during the loss of consciousness. Scale-free activity (as measured by power law exponent or detrended fluctuation analysis) is progressively reduced during the advancement of sleep stages N1 to N3 in the infraslow frequency range (Mitra et al., 2015; Tagliazucchi et al., 2013; Tagliazucchi & Laufs, 2014; Zhigalov et al., 2015). These studies observed progressive reduction in infraslow scale-free activity globally as well as in specific networks such as a DMN (including midline regions) and an attention network (including the lateral frontoparietal regions).

Whereas fMRI measures infraslow frequency range (<0.1 Hz), EEG usually targets higher-frequency ranges (1–180 Hz). Interestingly, misbalance between lower and higher frequencies, for example, stronger delta (1–4 Hz) with weaker beta and gamma (20–60 Hz), was found in unresponsive wakefulness state (UWS) and anesthesia (Lewis et al., 2012; Purdon et al., 2013; Sarà et al., 2011; Sitt et al., 2014). Studies on anesthesia also showed abnormal coupling of the ongoing phase in slow frequencies (0.01–1Hz) to either spiking rates (Lewis et al., 2012) or faster frequencies like alpha during the loss of consciousness (Mukamel et al., 2014).

Notably, an elegant EEG study by Purdon et al. (2013) showed that alpha amplitudes were maximal at low-frequency peaks during anesthetic-induced unconsciousness, whereas this relation reversed during consciousness and transition period to unconsciousness. Moreover, the phase–amplitude coupling and thus CFC predicted recovery of consciousness (Purdon et al., 2013). Taken together, the findings show that the temporal and spatial organization of the brain’s spontaneous activity by LRTS, CFC, scale-free, and small-world is central for the level/state of consciousness. Temporal nestedness of neural activity may thus not only organize and structure our brain’s spontaneous activity but also yield the level/state of consciousness (figure 7.1a).

Figure 7.1 Overview of different spatiotemporal mechanisms and the different dimensions of consciousness.

Let us now consider nestedness of neural activity on the spatial side. Barttfeld et al. (2015) showed reduced small-world organization in monkey anesthesia. Analogous findings were observed in human subjects (Uehara et al., 2014). It has been shown that dynamic functional connectivity was reduced in human anesthesia that resembled structural connectivity whereas in the awake state both structural and functional connectivity diverged transiently in specifically sensory regions (Tagliazucchi et al., 2016). Liu et al. (2014) observed reduced functional connectivity in both anesthesia and unresponsive wakefulness state (UWS); however, only UWS showed decreased scale-free properties while the latter were maintained in anesthesia.

Empirical Findings IIb Spatiotemporal Nestedness—Neural Predispositions of the Level/State of Consciousness

How can spatiotemporal nestedness of neural activity account for the level/state of consciousness? Spatiotemporal nestedness is a global feature of neural activity spanning across different time scales, frequencies, and regions or networks. Temporally, it refers to the integration or coupling across infraslow, slow, and fast frequencies, whereas spatially different regions/networks are integrated and organized in terms of small-world properties. Hence, spatiotemporal integration of different temporal and spatial scales allows constituting what we have described as spatiotemporal nestedness of neural activity.

Analogously, the level or state of consciousness can be considered a global feature that integrates and operates across different intrinsic temporal and spatial scales. For instance, the level/state of consciousness remains continuous across both short and longer time intervals, and this applies also to proximal and distal spatial environment. Psychologically, the level/state of consciousness may thus include different time and space scales that are nested within each other and may operate in a scale-free way. As such, the level/state of consciousness may well correspond to the spatiotemporal integration of the brain’s global integration of temporal and spatial dimensions. I propose that integration of different temporal and spatial scales is a central mechanism for constituting spatiotemporal nestedness in the level/state of consciousness.

In other words, I propose a correspondence of spatiotemporal scales between the brain’s spontaneous activity and the level/state of consciousness: the degree to which different temporal and spatial scales/ranges are integrated may correspond to the degree of temporal and spatial continuity of the level/state of consciousness. One may consequently hypothesize that fluctuations in the degree of spatiotemporal nestedness of the brain’s neural activity may correspond to analogous fluctuations in our level/state of consciousness across time and space. Accordingly, I propose a spatiotemporal correspondence between neural activity and the level/state of consciousness.

Moreover, it should be noted that the level/state of consciousness is not about specific contents but concerns “conscious experiences in their entirety, irrespective of their specific contents” (Koch et al., 2016). I maintain that spatiotemporal nestedness represents the brain’s neural activity in its entirety irrespective of specific contents. Therefore, spatiotemporal nestedness of the brain’s spontaneous activity is a necessary condition of possible consciousness—a neural predisposition of consciousness (NPC) (Northoff & Heiss, 2015).

Note that we here explicitly refer to the global spatiotemporal organization or structure of the brain’s spontaneous activity rather than to global activity or global metabolism (Schölvinck et al., 2010; Shulman et al., 2009) per se (or to a specific neural network such as the DMN). Global activity or metabolism may well be present without a specific spatiotemporal structure. We propose that it is the latter, the spatiotemporal structure that constitutes the degree of fractal and scale-free organization of the brain’s spontaneous activity, rather than the mere level of global activity or metabolism by itself, that is central for the level or state of consciousness. However, it should be noted that a sufficient metabolism level and thus energy supply may be necessary to constitute complex temporal and spatial structure—what we have termed spatiotemporal nestedness—in the brain’s neural activity. This may explain the findings, in specifically in UWRS, that the degree of glucose metabolism is usually the best predictor of the level/state of consciousness in these patients (Stender et al., 2014).

In sum, the studies just discussed demonstrate the central rule of LRTC, CFC, and small-world organization for the level/state of consciousness: LRTC, CFC, and small-world organization are spatiotemporal mechanisms that constitute the spontaneous activity’s structure as spatiotemporally nested. The level/state of consciousness is consequently a spatiotemporal phenomenon that can be traced back to the spatiotemporal nestedness of the brain’s spontaneous activity.

Let us rephrase the relation between neural activity and the level/state of consciousness. I propose that the spatiotemporal nestedness of the brain’s global neural activity is directly manifested in corresponding global features in consciousness, that is, its spatiotemporal nestedness. Consciousness is therefore as scale-free, as cross-frequency coupled, and as small-world organized as the spatiotemporal nestedness of the brain’s neural activity. If the brain’s neural activity is no longer scale-free, cross-frequency coupled, and small-world organized, as in anesthesia, sleep, and unresponsive wakefulness, then consciousness is lost (as its spatiotemporal nestedness is lost). Therefore, I consider spatiotemporal nestedness of the brain’s neural activity to be a distinct neural predisposition of the level/state of consciousness (NPC) (figure 7.2).

Figure 7.2 Temporal nestedness in the brain’s spontaneous activity and its relation to consciousness. (A) Measures of temporal structure in the brain’s spontaneous activity. (B) Changes in cross-frequency coupling during the loss of consciousness.

Empirical Findings IIc: Spatiotemporal Fragmentation and Isolation—Loss of the Level/State of Consciousness

How can we demonstrate in more detail that spatiotemporal nestedness of neural activity predisposes the level/state of consciousness? In a landmark study Lewis (Lewis et al., 2012) simultaneously investigated spiking in single unit recording and local field potentials (LFP) in three subjects with epilepsy during the loss of consciousness when undergoing anesthesia with propofol (see also Mukamel et al., 2014; Mukamel et al., 2011; Purdon et al., 2013; Purdon, Sampson, Pavone, & Brown, 2015).

During the application of increasing propofol dosage, subjects were stimulated with an auditory task during which they heard their own name (every 4 s) after which they had to respond by clicking a button. The loss of consciousness (LOC) was defined as the time period from 1s before the first missed stimulus to the second missed stimulus (in a sequence) amounting to a 5-s period (1 s + 4 s as interstimulus interval). This allowed Lewis et al. (2012) to compare the time period before the loss of consciousness (pre-LOC) with the time period after the loss of consciousness (post-LOC).

Let us start with a consideration of the spike rates. The spike rates decreased after 0–30 s of LOC with decreases of 81 percent to 92 percent when compared to pre-LOC. However, after around 4 min into the post-LOC state, the spike rates recovered and even increased (30%) when compared to pre-LOC: the spikes occurred in short periods in a highly dense way interrupted by rather long periods of total silence, or, total suppression. Hence spiking is preserved even when consciousness is lost; however, the firing pattern, that is, its temporal structure, changed showing long periods of silence.

What does this tell us about the LFP? The data showed a clear increase in the power of the slow oscillation (0.1–1 Hz), that is, slow cortical potentials (SCP) in post-LOC, which, unlike the spiking rates, remained stable during the entire period (5 min.) during which consciousness was lost. Other frequencies such as delta (increase), theta (decrease), alpha (increase), and gamma (increase) also changed but did not remain stable throughout the whole 5-min period when consciousness was lost. The authors conclude that the increase in the power of lower frequencies, the SCP, indicates the beginning of the loss of consciousness. In contrast, the power in higher frequencies was not directly related to the loss of consciousness (due to their instability in the period when consciousness was lost).

How are the LFP and especially the SCP related to the spiking rates? Lewis et al. (2012) observed that after LOC the spiking rate was significantly coupled to the phase of the SCP: 46.9 percent of the spikes from all recording units occurred near the trough (indexing high excitability as distinguished from the low excitability of the peak) in the phase of the slow oscillation (0.1–1 Hz). Most interestingly, such increased phase-spike coupling developed within seconds (–2.5–7.5 s) of LOC onset and may therefore be regarded as an index of LOC (Lewis et al., 2012).

The increase in phase-spiking coupling during LOC means that the firing is condensed to certain periods, that is, the trough, while the remaining periods, the descending, ascending, and peak parts of the slow oscillations’ cycle durations, did not show any firing any longer. Lewis et al. (2012) therefore speak of “on- and off-states” in firing rates and consequently of “temporal fragmentation” during LOC. In contrast, spiking is not as strongly entrained by the SCP phase when consciousness is still present: neurons fire here in all phases of the ongoing SCP including peak and trough as well as in ascending and descending phases.

Lewis et al. (2012) also investigated the relation between different channels and thus the way that the temporal dynamics translates into spatial distribution. For that, Lewis et al. (2012) calculated a phase-locking factor (PLF) between two oscillations in near and distant channels. Whereas the PLF for near channels was the same for pre- and post-LOC, it decreased proportionally with distance: the more distant the channel, the more variable the phase offsets of the respective channels. The same held analogously for the phase-spiking rates: the more distant spiking rates during post-LOC were no longer as strongly associated with a specific part of the ongoing phase, for example, trough, as the local firing rates. The local spatiotemporal dynamics is thus preserved, whereas the more distant or global spatiotemporal dynamics is broken down or fragmented—this suggests impaired communication between distant regions and thus spatial fragmentation with spatial isolation

In sum, the data suggest that the long-distance coupling of the slower frequencies’ phase to the faster frequencies’ amplitude, that is CFC, is disrupted implying “spatial fragmentation” of neural activity. Given the implications of these data, one can speak of spatiotemporal fragmentation with the loss of spatiotemporal nestedness, which is replaced by spatiotemporal isolation of neural activity during the unconscious state. Spatiotemporal isolation leads to the loss of temporal continuity of neural activity, which entails the loss of the “subjective feeling of continuity” (Tagliazzuchi et al., 2016, p. 10), indexing the breakdown of the level/state of consciousness.

How can we illustrate the importance of temporal continuity? The loss of temporal continuity in both neural activity and level/state of consciousness can be compared to the loss of continuity when one takes out two to three Russian dolls that are nested and contained within each other. Whereas from the outside everything looks the same (as the largest doll is preserved), the inner configuration with the degree of nestedness is altered once two to three of the smaller Russian dolls are removed. In the same way that the inner spatial continuity is no longer preserved in a nest of Russian dolls, the loss of temporal continuity on the neuronal level amounts to temporal fragmentation with the subsequent loss of temporal continuity, which, in turn, leads to the breakdown of the level/state of consciousness.

Spatiotemporal Model Ia: Spatiotemporal Integration versus Content-Based Integration

How can spatiotemporal integration account for the level/state of consciousness? The data of Lewis et al. (2012) show that the disruption of long-distance (cross-regional) temporal integration, that is, temporal fragmentation, leads to the loss of consciousness. Since temporal fragmentation of neural activity is accompanied by its spatial fragmentation, I speak of spatiotemporal fragmentation. The fact that spatiotemporal fragmentation features the loss of consciousness suggests a central role of spatiotemporal integration for consciousness. How can we determine spatiotemporal integration in more detail and distinguish it from other forms of integration?

First, we must consider exactly what is meant here by “integration.” One can describe different forms of integration, such as multisensory integration, perceptual integration, semantic integration, cognitive integration, and formal mathematical integration (see Mudrik, Faivre, & Koch, 2014, for an excellent overview). These forms of integration implicate the neural processing of different stimuli (or contents) and how they are related and linked together as in multisensory integration, perceptual integration, semantic integration, and cognitive integration. Moreover, there is also a purely formal form of integration, namely, mathematical integration (Mudrik et al., 2014).

What is the common denominator among these different forms of integration? They all, more or less, concern contents in their various facets and modalities. Different contents, such as sensory, cognitive, affective social, and others, are integrated with each other. One can therefore speak of content-based integration. Note that here I use the notion of “content” in a purely empirical way (rather than in a conceptual sense) and, in that sense, in a rather wide way—content-based integration concerns contents we can observe that can include sensory, affective, motor, cognitive, social, and additional contents. Hence, my empirical notion of contents is not restricted to cognitive contents (see part II in chapter 6).

How does such content-based integration stand in relation to the concept of integration as in spatiotemporal integration? Instead of integrating contents, spatiotemporal integration concerns the linkage of space and time by coupling different spatiotemporal scales or ranges in the brain’s neural activity. For instance, the temporal features of the spontaneous activity such as the fluctuations in the different frequencies are integrated as resulting in CFC and scale-free activity. Analogously, the spatial features in the brain’s neural activity such as the different regions are integrated in terms of cross-regional CFC, functional connectivity, and small-world properties.

Taken together, spatiotemporal integration is about the integration of different temporal and spatial scales or ranges rather than different contents. Therefore, I speak of spatiotemporal integration and distinguish it from content-based integration. I now posit that spatiotemporal integration in this sense characterizes the brain’s neural activity: this can be described by its spatiotemporal nestedness and be measured by LRTC, CFC, and small-world properties (and others).

Most importantly, I consider spatiotemporal integration of the different temporal and spatial scales within the brain’s neural activity central for consciousness: spatiotemporal integration allows for spatiotemporal nestedness, which can be considered a neural predisposition of the level/state of consciousness. The level/state of consciousness is thus based on spatiotemporal features like integration and nestedness rather than specific contents. Such spatiotemporal view is well compatible with the description of “conscious experiences in their entirety, irrespective of their specific contents” as expressed by Koch et al. (2016).

Spatiotemporal Model Ib: Spatiotemporal Integration without Content-Based Integration

The advocate of content-based integration may now want to argue that contents are associated with different spatiotemporal features. Therefore, the distinction between spatiotemporal and content-based integration will ultimately collapse and will no longer make a difference with regard to consciousness. Spatiotemporal integration is content-based integration, and, for that reason, consciousness is ultimately based on contents and their integration. I reject this argument since it neglects the difference and possible dissociation between spatiotemporal and content-based integration.

How are spatiotemporal and content-based integration related to each other? The advocate of content-based integration assumes that it is a necessary if not sufficient condition of spatiotemporal integration: without content-based integration, spatiotemporal integration and consequently consciousness remain impossible. I argue for the reverse situation. Specifically, I argue first that spatiotemporal integration can occur in the absence of content-based integration; and second, that content-based integration is based on and presupposes spatiotemporal integration. This amounts to the assumption that spatiotemporal integration is a necessary condition of content-based integration, whereas the latter is not a necessary condition of the former.

Can spatiotemporal integration occur in the absence of content-based integration? In that case, one would expect that spatiotemporal integration of neural activity could occur in the absence of and thus remains independent of any specific stimuli. That is, for instance, the case in the resting state in which no specific stimuli or tasks are applied to probe the brain’s stimulus-induced or task-evoked activity. The above described measures of LRTC, CFC, and small-world properties are indeed obtained in the resting state during the absence of stimuli or tasks. One can therefore assume that spatiotemporal integration of neural activity occurs in the resting state and remains thus independent of stimuli and their specific contents. Spatiotemporal integration can thus occur in the absence of content-based integration.

The advocate of content-based integration may now want to argue that even in the resting state plenty of stimuli are present. One can imagine, one has spontaneous thoughts, and there is continuous interoceptive input—there is thus no stimulus-free state. This implies that, even if present, spatiotemporal integration cannot be separated from content-based integration—that makes the argument of the independence of the former from the latter futile. The advocate is right. There is indeed continuous stimulus input and thus content-based integration even in the resting state.

However, there are extreme states where the balance between spatiotemporal and content-based integration is shifted toward the former at the expense of the latter. That is, for instance, the situation in meditation. In that case, one detaches the brain’s neural activity from the continuous stimulus input and their contents, that is, one’s own cognition, and, in extreme cases, also from one’s own body’s input (Tang & Northoff, 2017 Tang, Holzel, & Posner, 2015). The data suggest that spatiotemporal integration of the brain’s spontaneous activity remains and becomes probably even stronger in meditation during the absence of content-based integration (Tang, Holzel, & Posner, 2015, Tang & Northoff, 2017). Accordingly, although not fully detailed here, meditation can be considered an empirical example of spatiotemporal integration without or with minimal content-based integration.

Yet another example of spatiotemporal integration without content-based integration is that of psychiatric disorders. Depressed or manic patients suffering from bipolar disorder often show abnormally slow (depression) or fast (mania) temporal integration that is not accompanied by corresponding contents. They thus experience changes in their inner time consciousness without any related contents (Northoff et al., 2017). Taking both meditation and psychiatric disorders together, I argue that content-based integration is a not a necessary condition of spatiotemporal integration. This allows me to reject the argument that spatiotemporal integration is based on content-based integration or, in a weaker version, cannot be separated from contents.

Spatiotemporal Model Ic: Spatiotemporal Integration as Necessary Condition of Content-Based Integration

Can we consider spatiotemporal integration as a necessary condition of content-based integration? I argue that that is indeed the case and can be illustrated by the examples of multisensory integration (see chapter 10 in Northoff, 2014a, for details; see also Ferri et al., 2015; Stein et al. 2009) and temporal coding. Multisensory integration describes the integration between different sensory stimuli, that is, cross-modal stimuli. Such integration between different sensory stimuli follows certain mechanisms or principles. One such mechanism is the spatial coincidence of the two sensory stimuli, which enables and facilitates their integration: if the two cross-modal stimuli coincide at the same point in space, as for instance in a particular cell population or region, their likelihood of being integrated is much higher than when they do not spatially coincide and are processed in different cells or regions (see Stein et al., 2009; chapter 10 in Northoff, 2014a).

The same holds analogously for temporal coincidence: if the two sensory stimuli temporally coincide and thus occur at the same point in time, they can be much better integrated with each other than when they occur at different points in time. This makes it clear that multisensory stimuli are integrated with each other on the basis of their underlying spatial and temporal features, that is, their spatial and temporal coincidence (as based ultimately on probability distributions) (Northoff, 2014a). Spatiotemporal integration thus underlies and, at the same time, makes possible multisensory integration—the integration between different stimuli and their respective contents is defined by the integration of their underlying spatial and temporal features.

Spatiotemporal integration also underlies what can be described as “temporal coding” that describes the processing of information on the basis of temporal features. (Jensen et al., 2014). Gamma frequency (30–40 Hz) shows shorter cycle duration than alpha frequency (8–12 Hz). Presupposing CFC from alpha phase to gamma amplitude allows us to temporally segment the content that is processed via neural excitation according to the cycle duration of the frequencies: if the phase of the longer alpha cycle duration (100 ms) is coupled to and thereby allows excitation of the gamma amplitude, the contents may be temporally segmented according to the gamma cycles (10–30 ms).

Jensen et al. (2014) therefore speak of “phase coding” or “temporal coding.” The integration of contents, that is, which stimuli are integrated into contents and which ones are excluded or segregated, depends on the ongoing phase cycles of gamma and alpha and their CFC. The content is thus integrated on the basis of the temporal features of alpha and gamma—with the result being Jensen’s temporal coding. This example clearly reveals how the temporal (and possibly spatial) features of the brain’s neural activity allow for integrating contents—spatiotemporal integration with temporal and spatial coding may thus underlie content-based integration.

Because temporal and spatial coding allows for integrating contents, I propose that spatiotemporal integration is the most basic and fundamental form of integration. It already occurs in the spontaneous activity itself and therefore exists prior to stimulus-induced activity including its respective content. Moreover, it concerns the spatial and temporal features rather than the stimuli and their contents by themselves—spatiotemporal integration is therefore more basic than the various forms of integration involving stimuli and contents.

How is spatiotemporal integration related to consciousness? Spatiotemporal integration occurs and operates prior to and independent of consciousness. It occurs automatically and by default—spatiotemporal integration is an in-built mechanism. We are not and, put more strongly, we cannot become aware or conscious of such integration between different frequencies and regions in our brain’s neural activity. Nor can we become aware of corresponding integration between different stimuli or contents in our consciousness. Instead, we can only access the ready-made result, that is, spatiotemporal nestedness, that becomes manifest at the level/state of actual consciousness.

Spatiotemporal Model IIa: Argument of Nonspecificity—Specificity of Spatiotemporal Mechanisms for Brain and Consciousness

The critic may now to argue that although spatiotemporal mechanisms, including integration and expansion, are empirically grounded, they remain unspecific when it comes to consciousness. The very same spatiotemporal mechanisms can operate during both consciousness and unconsciousness. This holds especially given that any neuronal mechanisms of the brain is based on the temporal and spatial features of its neural activity regardless of whether it is associated with consciousness. The claim of the argument is thus that spatiotemporal mechanisms do not allow us to distinguish between conscious and unconscious states and therefore remain unspecific. We may thus put forward such nonspecificity against the thesis that spatiotemporal mechanisms characterize consciousness: I therefore speak of an argument of nonspecificity.

What are we to make of the argument of nonspecificity? I reject this argument. True, any neural activity in the brain is spatial and temporal, as it takes place in time and space. Therefore, any neural activity in the brain will be spatial and temporal by default—time and space in such a general sense remain indeed too unspecific to distinguish consciousness and unconsciousness. For that reason, the mere temporal and spatial nature of the brain’s neural activity does not permit us to distinguish between spatiotemporal mechanisms involved in consciousness as distinguished from unconsciousness.

However, the argument of nonspecificity neglects the fact that the brain can construct the temporal and spatial features of its neural activity in different ways. For instance, the brain’s neural activity may show spatiotemporal nestedness with strong LRTC, CFC, and small-world properties. Yet at the same time, neural activity may also show spatiotemporal fragmentation rather than nestedness (see the section on time, space, and the brain, above). Moreover, concerning stimulus-induced activity, there may either be spatiotemporal expansion or, as I propose, spatiotemporal constriction (as in the TMS-EEG presented by Massimini et al. 2005, 2007; Casal et al., 2013; see part II in chapter 4 for details).

These examples demonstrate that the brain’s spontaneous and stimulus-induced activity can be constructed in different ways. These different ways may be distinguished from each other on temporal and spatial grounds—they are thus far from being unspecific: the brain’s neural activity can construct the temporal and spatial features of its neural activity in distinctive ways that lead to different organization, for example spatiotemporal nestedness versus fragmentation and spatiotemporal expansion versus constriction. We may therefore confidently rebut the argument of nonspecificity with respect to the spatiotemporal characterization of the brain’s neural activity: the spatiotemporal mechanisms allow for a specific organization of the brain’s neural activity as distinguished from others.

Moreover, the data suggest that such spatiotemporal organization of neural activity allows distinguishing between consciousness and unconsciousness. Based on the data, I propose that spatiotemporal organization of neural activity in a specific way (in this case, spatiotemporal nestedness and expansion rather than fragmentation and constriction) distinguishes consciousness from unconsciousness. Therefore, I rebut the argument of nonspecificity with respect to the spatiotemporal characterization of consciousness and its relation to the brain: the underlying spatiotemporal mechanisms are not nonspecific but rather specific for consciousness, as distinguished from those underlying unconsciousness.

Spatiotemporal Model IIb: Argument of Nonspecificity—Spatiotemporal Mechanisms and Models of Brain

How can we further substantiate the specific nature of the postulated spatiotemporal mechanisms? We have suggested different models of brain including the spectrum model (chapter 1) and the interaction model (chapter 2). To further support the specificity of the supposed spatiotemporal mechanisms, we may also want to link them to the models of brain.

Let us start with a consideration of the spectrum model of brain as presented in chapter 1.

The spectrum model argues for a continuum between passive and active features in neural activity; it therefore constitutes a “hybrid” by default, that is, a mixture of both stimulus-induced and spontaneous activity (chapter 1). Spatiotemporal expansion as associated with consciousness is well compatible with such a model: higher degrees of spatiotemporal expansion make stimulus-induced activity more hybrid and shift it more toward the active pole of the spectrum. The opposite is the case in unconsciousness: high degrees of spatiotemporal constriction lead to less hybrid stimulus-induced activity (as less impacted by the spontaneous activity), which shifts it more toward the passive pole of the spectrum.

Analogously, the same holds true for spatiotemporal nestedness. High degrees of LRTC, CFC, and small-world properties shift the brain’s neural activity toward the “active” pole of the spectrum. In contrast, spatiotemporal isolation with low degrees of LRTC, CFC, and small-world properties shifts the brain’s neural activity toward the opposite pole, the “passive” pole of the spectrum. The neural distinction between spatiotemporal nestedness and isolation, including their association with consciousness and unconsciousness, is thus well compatible with the spectrum model of brain.

We can make an analogous case for the interaction model of brain that postulates a continuum of different degrees of nonadditive interaction between spontaneous activity and stimuli (chapter 2). Data show that spatiotemporal expansion is related to a higher degree of nonadditive rest-stimulus interaction (Huang et al., 2015), which, in turn, seems to be sufficient for associating stimuli with consciousness (chapter 5). The degree of nonadditivity in rest-stimulus interaction is rather low in cases of spatiotemporal constriction as during unconsciousness. Hence, the distinction between consciousness and unconsciousness by spatiotemporal expansion and constriction is well compatible with the interaction model of brain.

In conclusion, I reject the argument of nonspecificity against the spatiotemporal model of consciousness by arguing for what I describe as spatiotemporal specificity. Yes, the brain’s neural activity in general is indeed both spatial and temporal. However, the advocate of this argument ignores the fact that those very same spatiotemporal features of the brain’s neural activity can be constructed in different ways as related to different spatiotemporal mechanisms. The different spatiotetmporal mechanisme are central for distinguishing between the presence and absence of consciousness. Accordingly, based on the empirical data reviewed in this section, I argue for spatiotemporal specificity in the relationship between brain and consciousness.

Finally, we should note that the rejection of the argument of nonspecificity concerns the phenomenal features of consciousness and thus noncognitive consciousness as presented in the spatiotemporal model of consciousness. By contrast, we leave open here whether the argument may apply to “cognitive consciousness” as specified in the GNWT. The spatiotemporal characterization, as suggested here, may indeed remain too unspecific when it comes to the cognitive features of consciousness.

Spatiotemporal Model of Consciousness II: Spatiotemporal Expansion—Neural Correlate of Consciousness and Neuronal-Phenomenal Correspondence

Empirical Findings Ia: Stimulus-Induced Activity—Amplitude Increase and Trial-to-Trial Variability Reduction

We have so far discussed that the spatiotemporal nestedness of the brain’s spontaneous activity provides a neural predisposition of the level/state of consciousness or NPC. The spatiotemporal alignment of the brain’s spontaneous and/or prestimulus activity to single stimuli and long-term stimulus sequences in our body and the world is considered to be a neural prerequisite of consciousness, what we have previously designated as preNCC. This leaves open the neural correlates of consciousness, what we have previously designated as NCC—the neural mechanisms that are sufficient to associate specific contents with consciousness.

How, we may ask, is the stimulus processed in neural terms so that it can be associated with consciousness? Data have revealed that the amplitude of stimulus-evoked neural activity can be considered a marker of consciousness: the higher the amplitude in response to the stimulus, the more likely the stimulus will be associated with consciousness (Koch et al., 2016; Tsuchiya et al., 2015). In addition, data from various studies have shown that, compared to unconsciousness processing, stimulus-induced activity during consciousness lasts longer and is spatially more extended. This has been reviewed extensively in recent papers (Dehaene et al., 2014; Koch et al., 2016); for this reason, we only highlight here some results on temporal duration and spatial extension (figure 7.3).

Figure 7.3 Spatiotemporal expansion of neural activity.

Li et al. (2014) measured slow cortical potentials in MEG during presentation of near-threshold visual stimuli. The MEG results showed long-lasting event-related magnetic fields (ERMF) between 300 ms and 2–3 s poststimulus during seen trials when compared to unseen ones. The long-lasting ERMF do not resemble oscillations but slow DC-type drift (a slow change in the power of the frequency over longer stretches of time), making them likely reflect slow cortical potentials in the slow frequency range between 0.1 and 5 Hz. Interestingly, the long lasting ERMF were specific for subjective awareness, the seen versus the unseen, whereas they appeared neither in objective performance, for example, as distinction between correct and incorrect trials, nor in confidence judgments. The long-lasting ERMF changes during seen trials were accompanied by widespread activity changes in temporal and frontoparietal cortices. One can therefore suppose that slow cortical potentials shape stimulus-induced activity and its association with consciousness; this view is further supported by the findings revealed in phase and power analysis discussed in the same study (see Li et al., 2014).

The central role of spatiotemporal expansion is impressively demonstrated in the TMS-EEG experiments by the group around Massimini (Casali et al., 2013; Massimini et al., 2010). They applied the same TMS pulse (in premotor and parietal cortical regions) during both conscious and unconscious states in UWRS, anesthesia, and sleep. The degree of spatiotemporal expansion of TMS-induced activity varied considerably between conscious and unconscious states. During the conscious state, the TMS-pulse induced spatially extended activity for long durations. In contrast, both temporal duration and spatial extension of TMS-induced activity were extremely restricted during the unconscious state in anesthesia, UWRS, and sleep.

Why and how is it possible that the same TMS pulse leads to different spatiotemporal features of the stimulus-induced activities during conscious versus unconscious states? I posit that this is related to the spontaneous activity itself as well as its neural interaction with the TMS pulse, that is, the rest-stimulus (or rest-pulse) interaction (Huang, Zhang, Longtin, et al., 2017; Northoff et al., 2010). The fact that stimulus-induced activity expands during consciousness suggests that the stimulus can better suppress the ongoing spontaneous activity: the better the stimulus interacts with the spontaneous activity and suppresses its ongoing fluctuations, the more likely the stimulus can expand the temporal duration and spatial extension of the activity it induces. Such suppression of the ongoing spontaneous activity by the stimulus can be measured by trial-to-trial variability, which has been shown on both cellular (Churchland et al., 2010) and regional (Ferri et al., 2015; He, 2013; Huang, Zhang, Longtin, et al., 2017) levels of neural activity. The more the stimulus suppresses the variability of the ongoing spontaneous activity, the more (negative) interaction between spontaneous and stimulus-evoked activity (He, 2013; Huang, Zhang, Longtin, et al., 2017). Although the exact neural mechanisms of such suppression of spontaneous activity fluctuations by the stimulus remain unclear, I would hypothesize that the spatiotemporal expansion of stimulus-induced activity is related to the suppression of ongoing activity and rest-stimulus interactions.

Empirical Findings Ib: Spatiotemporal Expansion: Expansion versus Integration—Integrated Information Theory

How does the supposed spatiotemporal expansion of stimulus-induced activity as NCC stand in relation to integration and the integrated information (IIT) (Tononi et al., 2016; Tononi & Koch, 2015)? Stated simply, the main thesis of the IIT is that consciousness is based on information integration that has been mathematically formalized in the phi index, while, empirically, it has been operationalized by the perturbation complexity index (Casali et al., 2013). One should also note that the concept of information in the IIT does not refer to information in terms of specific contents as traditionally or commonly understood (Tononi & Koch, 2015). Rather, within the context of the IIT, information refers to “how a system of mechanisms in a state, through its cause-effect power, specifies a form (‘informs’ conceptual structure) in the space of possibilities” (Tononi & Koch, 2015, p. 8; emphasis added). Let us explicate that quote in spatiotemporal terms of the STC with further consideration focusing especially on those concepts highlighted by italics.

How do Tononi and Koch’s concepts of “form,” “space of possibilities,” and “state” stand in relation to spatiotemporal expansion advocated here? The data suggest that the stimulus both suppresses and enhances the neuronal features of the brain’s spontaneous activity. Spatiotemporal expansion is thus dependent on and ultimately based on spatiotemporal nestedness and alignment: without proper degrees of spatiotemporal nestedness and alignment, the stimulus will not be able to spatiotemporally expand its stimulus-induced activity. One may assume that the constellation of the various spatiotemporal mechanisms (which, as we predict, will be complemented by various other spatiotemporal mechanisms in the future) corresponds to what Tononi and Koch (2015, p. 8) describe as “system of mechanisms.” We suggest that such a “system of mechanisms” operates on a spatiotemporal platform, hence our suggestion of the various spatiotemporal mechanisms.

What exactly is meant by “a state”? I suggest that “a state” refers to the brain’s spontaneous activity and, more specifically, its degree of spatiotemporal nestedness. The various spatiotemporal mechanisms, as suggested here, operate within the space of the brain’s spontaneous activity and its spatiotemporal structure: both spatiotemporal alignment and spatiotemporal expansion are based and dependent on the spontaneous activity’s spatiotemporal nestedness. At the same time, however, the very same spatiotemporal nestedness is modulated by increasingly longer stimulus sequences; the “state,” that is, the spontaneous activity’s spatiotemporal nestedness as we say, is itself not fixed but, rather, highly dynamic and malleable.

How, then, may we conceive the concepts of “form (‘informs’ conceptual structure)” and “space of possibilities” as raised in IIT in the spatiotemporal context of STC? The STC suggests that spatiotemporal alignment of the brain’s neural activity to the spatiotemporal structure of body and world provides a spatiotemporally based form as background and third dimension (in addition to level/state and content) of consciousness. Therefore, we suggest that what the IIT describes as “form” may be traced to the virtual spatiotemporal structure or organization among brain, body, and world that constitutes the form we experience as the third dimension of consciousness. More specifically, what Tononi describes as “conceptual structure” may then be traced to what we describe as “form” characterized by spatiotemporal structure as ranging in a virtual probability-based way between world and brain (i.e., the world-brain relation).

The very same spatiotemporal structure that ranges among brain, body, and the world provides the neural prerequisite (or, to be more precise, a neuro-ecological prerequisite) of consciousness (preNCC). This spatiotemporal structure is probabilistic and renders certain ways of spatiotemporal expansion and stimulus-induced activity possible while it excludes others. If, for instance, the spontaneous activity’s various frequencies are already suppressed and nonreactive in the resting state, the stimulus will not be able to enhance them to expand its own stimulus-induced activity; this, in turn, makes it impossible to associate stimulus-induced activity with consciousness. Accordingly, what Tononi and Koch (2015) (see also Tononi et al., 2016) describe as a “space of possibilities” may find its more specific neuronal mechanisms in spatiotemporal nestedness as NPC and spatiotemporal alignment as preNCC. His “space of possibilities” is thus a space of possible spatiotemporal configurations.

In sum, I propose that what we term here the STC or spatiotemporal approach to consciousness is well compatible with the IIT. Integration as in IIT is supposed to occur within a temporal and spatial arena rather than on a sensory or cognitive basis. Moreover, we regard the concrete spatiotemporal determination of specific neuronal mechanisms, spatiotemporal mechanisms, involved in consciousness as complementary to the more abstract notion of information in IIT. Future investigation may want to apply some of the mathematical and operational measures of the IIT in the context of the presumed spatiotemporal mechanisms of STC.

Empirical Findings IIa: Stimulus-Induced Activity—Early versus Late

How can we investigate consciousness? The traditional way is to ask subjects whether they saw or heard something and then to make a judgment; this has recently been described as “report paradigm” (Tsuchiya et al., 2015). However, the judgment or report itself may introduce a cognitive component that as such may not belong to consciousness proper. The neural mechanisms underlying such a judgment or report may thus be a neural consequence of rather than a neural correlate of consciousness (see Aru, Bachmann, Singer, & Melloni, 2012; deGraaf et al., 2012; Li et al., 2014). For this reason, report paradigms have been contrasted with “no-report paradigms” where the subjects do not need to report or to give a judgment (Lamme, 2010a,b; Tsuchiya et al., 2015).

No-report paradigms reveal a different spatial and temporal pattern from report paradigms. Several studies have demonstrated that early components (such as P50 and N100) of stimulus-induced activity (from around 100 ms to 200–300 ms) indicate the presence and experience of a specific content in consciousness even if that very same content may not yet be accessible for subsequent reporting (Andersen, Pedersen, Sandberg, & Overgaard, 2015; Koch et al., 2016; Koivisto et al., 2016; Koivisto & Revensuo, 2010; Palva et al., 2005; Pitts, Metzler, & Hillyard, 2014; Pitts, Padwal, Fennelly, Martínez, & Hillyard, 2014; Rutiku, Martin, Bachmann, & Aru, 2015; Schurger, Sarigiannidis, Naccache, Sitt, & Dehaene, 2015; Tsuchiya et al., 2015). The presence of these electrophysiological markers of early stimulus-induced activity such as N100 is reduced if not completely absent in altered states of consciousness such as anesthesia, slow-wave sleep, and vegetative state (Bachmann & Hudetz, 2015; Koch et al., 2016; Purdon et al., 2013; Sitt et al., 2014; Schurger et al., 2015).

We now consider the spatial side. The report paradigms of Tsuchiya et al. show extensive involvement of the especially lateral prefrontal and parietal cortical regions (Tsuchiya et al., 2015). In contrast, as reviewed in detail in Tsuchiya et al. (2015) and in Koch et al. (2016), no-report paradigms do not show prefrontal-parietal recruitment but rather posterior cortical regions at the interface between parietal, occipital, and temporal cortex with a specific focus on higher sensory regions, or, “hot zones” as Koch (see Koch et al., 2016) calls them.

Moreover, we need to consider the distinction between medial and lateral prefrontal regions. Lateral prefrontal regions are recruited during judgment, in what may be termed report paradigms because they are related to various cognitive functions including working memory (Northoff et al., 2004; Tsuchiya et al., 2015). In contrast, due to the lower cognitive load as related to the absence of judgment, no-report paradigms lead to stronger involvement of medial prefrontal regions such as the ventromedial prefrontal cortex that present lower degrees of deactivation in fMRI (Northoff et al., 2004; Shulman et al., 1997a,b). This is well compatible with the involvement of the midline regions in various forms of spontaneous mental activity associated with consciousness such as spontaneous thoughts (Christoff et al., 2016), mental time travel with episodic simulation (Schacter et al., 2012), and self-related processing (Northoff, 2016d; Northoff et al., 2006).

What does the operational distinction between the report and no-report paradigms, and their different neuronal correlates, imply for our characterization of consciousness? The late event-related potentials such as the P300 and lateral prefronto-parietal regions seem to be related to the cognitive functions implicated in judgment of stimuli as conscious (see Silverstein et al., 2015, for providing evidence that the P3b may occur even during unconscious states). Late event-related potentials and lateral prefronto-parietal cortical activity may consequently be associated with awareness of the content of consciousness rather than with consciousness itself. We therefore postulate that these neuronal measures obtained in report paradigms reflect what has been conceptually described as “neural consequence of consciousness” (NCC con) rather than the NCC proper (Aru et al., 2012; deGraaf et al., 2012; Northoff, 2014b).

In contrast to report paradigms, no-report paradigms reveal earlier event-related potentials such as N100 and 200 as well as posterior cortical regions and/or cortical midline regions. I propose that these earlier spatiotemporal features of stimulus-induced activity reflect the NCC as detailed in the previous section. Taken in this sense, the NCC must be distinguished from the NCC con in both operational and neuronal terms.

Empirical Findings IIb: Cognitive Features of Consciousness—Global Neuronal Workspace Theory

Relying on report paradigms, the various findings supporting the global neuronal workspace theory (GNWT) show later components such as P300 and prefrontal cortical involvement (see Baars 2005; Dehaene et al., 2014; Dehaene & Changeux, 2011, for excellent reviews). We do not recapitulate here the various findings that support the GNWT, which have been reviewed in detail elsewhere (Dehaene & Changeux, 2011; Dehaene et al., 2014). Due to the impressive findings it has revealed, however, the question is not so much whether late prefrontal activity is related to consciousness in general but, rather, to what feature of consciousness it is related. That distinction is the focus of the discussion that follows.

The GNWT postulates that the environmental stimuli and their respective contents become globally available for cognition. Such globalizing and sharing presumably is made possible by the architecture of the brain, most especially the design of both the lateral prefrontal and parietal cortex as a “global workspace” where the different functional systems of the brain (specifically memory, evaluative/reward, attentional, motor, and perceptual systems) converge and overlap. We describe that confluence as spatial globalization of local-regional, stimulus-induced activity that is globalized throughout the whole brain including prefrontal and parietal cortex. Dehaene (Dehaene et al., 2014) and Moutard (Moutard, Dehaene, & Malach, 2015) characterize this as “non-linear ignition”: they suggest that the transition from merely local-regional to global prefrontal-parietal activity must be ignited by the stimulus in a nonlinear rather than merely a linear way. The exact neuronal mechanisms of such nonlinear ignition of lateral prefrontal-parietal cortical activity, however, remain unclear.

In addition to its spatial component, the GNWT also considers temporal measures such as late event-related potentials (P300), synchronization between regions, and high-frequency oscillations including gamma as central for consciousness (Dehaene & Changeux, 2011; Dehaene et al., 2014). Analogous to spatial globalization, such extension of neural activity to different temporal domains from the early sensory-based event-related potentials such as N100 (see Bachmann & Hudetz, 2015) to later event-related potentials and higher frequencies such as gamma may be described by the term temporal globalization. Such temporal globalization seems to be deficient in altered states of consciousness where the temporal extension to later event-related potentials (as in reduced P300) and higher frequencies (as in reduced gamma power) is no longer present (Koch et al., 2016; Sitt et al., 2014) (see figure 7.4).

Figure 7.4 Temporospatial globalization of neural activity and consciousness.

Which feature or aspect of consciousness is targeted by the GNWT and, more specifically, by its assumed spatiotemporal globalization of stimulus-induced activity? By relying on report paradigms, the GNWT indexes consciousness by accessing and reporting the contents of consciousness. The access and reporting of contents is experimentally reflected in that the participants are required to access and report the stimulus content by clicking a button with intent that is taken to represent an index of consciousness (Dehaene & Changeux, 2011; Dehaene et al., 2014). The lateral prefrontal-parietal cortex and the late components of stimulus-induced activity from 300 ms to 500 ms such as P300 (or even longer up to 800 ms) are strongly associated with cognitive functions such as selective attention, expectation, self-monitoring, and task planning and, most importantly, accessing and reporting the contents of consciousness (Koch et al., 2016; Tsuchiya et al., 2015, Andersen et al., 2016 Aru et al., 2012; deGraaf et al., 2012; Pitts, Metzler, et al., 2014; Pitts, Padwal, et al., 2014; Rutiku et al., 2015; Schurger et al. 2015; Tsuchiya et al., 2015).

Spatiotemporal globalization of stimulus-induced activity as required for the accessing and reporting of contents in consciousness must be distinguished from consciousness itself. This is supported by the data obtained using nonreport paradigms as discussed in the previous section. I therefore conceive of spatiotemporal globalization as the neural consequence of consciousness or NCC con (see Andersen et al., 2016 Aru et al., 2012; deGraaf, Hsieh, & Sack, 2012; Li et al., 2014; Overgaard & Fazekas, 2016; Tononi & Koch, 2015; Tsuchiya et al., 2015).

To summarize, we suggest that spatiotemporal globalization, as postulated in GNWT, allows for late and prefronto-parietal cortical involvement in stimulus-induced activity during consciousness. Late and prefronto-parietal cortical involvement is experimentally related to accessing and reporting of contents. Therefore, I propose that late prefronto-parietal activity and thus spatiotemporal globalization is neural consequence of the phenomenal features consciousness that occur prior to and independent of accessing and reporting contents. As it is related to the access and reporting of contents, spatiotemporal globalization must be distinguished from spatiotemporal expansion of stimulus-induced activity that, as we suggest, is more related to the phenomenal features of consciousness.

Spatiotemporal Model Ia: From Spatiotemporal Expansion to Phenomenal Features

We have so far discussed various neuronal mechanisms such as spatiotemporal expansion and globalization. This leaves open the phenomenal features of consciousness including their relation to these neuronal mechanisms. These subjects are our next focus of consideration.

What are the phenomenal features of consciousness and how are they neuronally instantiated? In essence, this is the central question of both neuroscience and the philosophy of consciousness. Phenomenal features refer to experience that is subjective rather than objective. Philosophers often describe phenomenal features by their qualitative character—from qualia as the “what it is like” of experience (Nagel, 1974). Other phenomenal features concern the directedness of experience toward a specific content, for example, intentionality (Searle, 2004), self-perceptiveness with first-person perspective, and others (Northoff, 2014b, 2016c,d).

Without going into detail, we recognize that these phenomenal features must be distinguished from the cognitive features of consciousness. Phenomenal features concern the experience of a content as conscious, whereas cognitive features allow one to access and subsequently report that very same content. The distinction between phenomenal and cognitive features of consciousness is more or less mirrored in the conceptual distinction between noncognitive and cognitive consciousness (Cerullo et al., 2015).

How can spatiotemporal expansion of stimulus-induced activity serve as a neural correlate of the phenomenal features of consciousness? Buzsáki contends that “perception goes beyond the stimulus” (Buzsáki, 2006). We now suggest that such “beyond” is central for the phenomenal features of consciousness and that it consists of spatiotemporal features. The stimulus itself can be characterized in temporal and spatial terms by its duration and extension. That contrasts with the experience or consciousness within which the very same stimulus usually lasts longer and extends beyond its mere physical duration and extension. For instance, even though they may be physically absent, we may still hear the last tones of a melody and may still visualize the last scene of an opera in a more temporally and spatially extended sense. In brief, the spatiotemporal features of the stimulus in consciousness thus expand beyond the ones featuring the stimulus in purely physical terms so that these features show both longer temporal duration and yield more distributed spatial extension.

From where does such spatiotemporal expansion beyond the stimulus’s own physical duration and extension in consciousness emerge? We tentatively suggest that it arises from or, put differently, is appended to the stimulus by the brain’s spontaneous activity. The brain’s spontaneous activity shows a large temporal and spatial scale or range that extends far beyond the range of the single stimulus. For instance, the stimulus may show a duration of 100 ms, which corresponds to alpha frequency or a duration of 1 s as mirroring delta frequency. The spontaneous activity, in contrast, includes a much wider variety of different frequencies ranging from infraslow frequencies to considerably faster ones.

This carries major implications for how the stimulus is processed by the brain’s spontaneous activity, the, “rest-stimulus interaction” (Northoff et al., 2010). When interacting with the spontaneous activity, the stimulus and its more limited spatiotemporal scale during its presentation interact with a much larger spatiotemporal range of the brain’s spontaneous activity. This, in turn, makes possible for the brain to integrate, nest, and contain the stimulus and its associated stimulus-induced activity within the spontaneous activity and its larger spatiotemporal scale. This, in turn, expands the stimulus’s own temporal duration and spatial extension beyond itself according to the brain’s intrinsic temporal duration and spatial extension of its spontaneous activity. Such nesting or embedding may, for instance, be manifested in the stimulus-induced modulation of the spontaneous activity’s scale-free activity, which allows expansion of the stimulus beyond its own original spatiotemporal scales.

In sum, I hypothesize that the degree of spatiotemporal expansion of the stimulus beyond its own or original spatiotemporal features (as during the stimulus’ presentation in the experimental situation) is closely related to the association of the stimulus/contents with the phenomenal features. Given such spatiotemporal expansion of the stimulus by the brain’s spontaneous activity, the resulting phenomenal features may by themselves be characterized as spatiotemporal in a virtual way: the degree of the stimulus’s spatiotemporal expansion on the neuronal level of rest–stimulus interaction may directly correspond to the experience of the spatiotemporal features of the stimulus in consciousness. Therefore, I postulate that consciousness and its phenomenal features are spatiotemporal features.

Spatiotemporal Model Ib: From Spatiotemporal Features to Neuronal-Phenomenal Correspondence

Given the central role of spatiotemporal expansion, one expects correspondence between the spatiotemporal features of the brain’s spontaneous activity as modulated by the stimulus on the one hand and the spatiotemporal features during the experience of the stimulus in consciousness on the other. I therefore speak of neuronal-phenomenal correspondence between phenomenal features and the spatiotemporally expanded stimulus-induced activity. Neuronal-phenomenal correspondence as described here yields “the neurophenomenal hypothesis” (Northoff, 2014b), although conceptually it may be described variously as “isomorphism” (Fell, 2004), “operational time-space” (Fingelkurts et al., 2013), or “identity” (Tononi et al., 2016).

Note that such neuronal-phenomenal correspondence exists between the brain’s neuronal activity and the phenomenal features of consciousness. In contrast, there is no such correspondence between the stimulus itself, that is, its physical features in terms of time and space and the phenomenal features of consciousness. There is thus a discrepancy between physical and phenomenal features of the stimulus—a physical-phenomenal discrepancy that one may say comes close to what we have described as a physical-neuronal discrepancy. There is thus a gap between physical and phenomenal features. This gap can be filled or bridged by the brain’s spontaneous activity and its spatiotemporal features.

More specifically, the brain’s spontaneous activity provides a spatiotemporal framework that allows us to associate phenomenal features with the otherwise purely physical stimulus. As they are based on the spontaneous activity’s spatiotemporal features, phenomenal features must by themselves be characterized as spatiotemporal. Time and space are here understood in the sense of “inner duration” and “inner extension,” as discussed above in the section on time, space, and brain. Presupposing such a sense of time and space, phenomenal features can be characterized as spatiotemporal—phenomenal features are spatiotemporal features.

Spatiotemporal features link the brain’s neuronal features to the phenomenal features of consciousness. The glue or “common currency” between brain and consciousness and thus between neuronal and phenomenal features consists in spatiotemporal features. Construing spatiotemporal features as common currency allows us to associate phenomenal features with the stimulus by processing and integrating the latter within the brain’s spontaneous activity.

Importantly, such integration allows for expanding the spatiotemporal features of the stimulus by the spontaneous activity’s spatiotemporal structure; if such spatiotemporal expansion is sufficient and goes beyond the stimulus itself, the resulting stimulus-induced activity can be associated with consciousness and its phenomenal features. The phenomenal features of consciousness can consequently be characterized by spatiotemporal features that correspond to (and ultimately are shared with) the spatiotemporal features of the brain’s spontaneous activity and the degree to which the latter expands the stimulus’s spatiotemporal features. Accordingly, spatiotemporal expansion can be considered the neuronal mechanism that underlies neuronal-phenomenal correspondence.

Note that the concept of neuronal-phenomenal correspondence is a bridge concept between neuronal, that is, empirical, and phenomenal, that is, experiential, domains. The concept is not purely empirical, as phenomenal features cannot be observed from the third-person perspective in the same way we observe neuronal states in neuroscience. At the same time, neuronal-phenomenal correspondence is not a fully phenomenal concept, either, as such correspondence cannot be experienced in consciousness. Hence, I consider neuronal-phenomenal correspondence a truly “neuro-phenomenal concept” (Northoff, 2014b).

Finally, note that neuronal-phenomenal correspondence is also not an ontological concept. The concept only claims a correspondence between neuronal and phenomenal states with respect to spatiotemporal features. As such, neuronal-phenomenal correspondence does not imply any assumption about whether phenomenal features exist independent of neuronal features. It also implies nothing about whether the existence of phenomenal features and thus consciousness can be reduced to the existence of neuronal states and the brain. Accordingly, neuronal-phenomenal correspondence remains neutral to any such ontological claims.

Neuronal-phenomenal correspondence may nevertheless carry im­portant implications for the ontological domain. As stated, neuronal-phenomenal correspondence only describes that neuronal and phenomenal features show corresponding spatiotemporal features; this remains independent of any claims of the existence underlying both neuronal and phenomenal features. One may extend the claim of correspondence to the ontological domain, however. In that case, one does not only claim a spatiotemporal correspondence but, what is much stronger, that neuronal and phenomenal features share the same underlying spatiotemporal features. The existence and reality of phenomenal and neuronal features is consequently based on and traced to spatiotemporal features. Such ontological extension of neuronal-phenomenal correspondence requires an ontology that considers space and time as the basic units of existence and reality. I claim that both consciousness and brain, and thus neuronal and phenomenal features, can indeed by characterized such a spatiotemporal ontology that describes their commonly underlying existence and reality. Such a spatiotemporal ontology will be developed in the third part of this book.

Spatiotemporal Model IIa: Argument of Triviality—The Empirical versus the Conceptual-Logical

We may now revisit and reject what I have referred to as the argument of triviality. The argument of triviality claims that the characterization of both brain and consciousness by spatiotemporal features is trivial. Briefly, the argument posits that characterizing the brain’s neural activity by spatiotemporal features is trivial since neural activity is intrinsically both spatial, that is, it involves different regions, and temporal, it covers different frequencies. The brain’s neural activity is consequently spatiotemporal by default because to imply otherwise, for example in the absence of any spatiotemporal features, would be to accept absence of any neural activity whatsoever—the brain would no longer be a brain and thus simply remain absent.

Characterizing the brain’s neural activity by spatiotemporal features may consequently be considered trivial because this characterization does not say something novel or add anything to what we already know about the brain and its neural activity. By analogy, the same obvious characterization also applies to consciousness and its spatiotemporal characterization: any behavior including consciousness and other mental states is temporal and spatial by the very nature of the underlying brain and its neural activity. Therefore, much like the spatiotemporal characterization of the brain’s neural activity, the spatiotemporal model of consciousness is also trivial. This is what I call the argument of triviality.

One may now wonder how the argument of triviality may be distinguished from the argument of nonspecificity. The argument of nonspecificity concerns the specific versus the unspecific nature of spatiotemporal mechanisms, that is, whether the spatiotemporal mechanisms are associated with specific patterns and organization of the brain’s neural activity as well as with consciousness as distinguished from unconsciousness (see above in the first part on the spatiotemporal model of consciousness). The argument of nonspecificity can thus be characterized as an empirical argument. It is one that concerns the question for the empirical specificity of spatiotemporal mechanisms.

The argument of triviality, in contrast to the argument of nonspecificity, extends beyond spatiotemporal mechanisms and raises a deeper and more basic question: Does the characterization of brain and consciousness by time and space in general tell us anything at all? The argument is thus no longer merely empirical, as is the argument of unspecificity, but, rather, it entails a strong conceptual–logical (and ultimately also ontological) dimension: it raises the question of the characterization of the basic ingredients underlying the spatiotemporal mechanisms. Therefore, I now provide a rejection of the argument of triviality on conceptual-logical grounds that complements my empirical rejection of the same argument in chapter 4. To address the argument of triviality, we therefore need to venture from the empirical to more conceptual–logical grounds.

Spatiotemporal Model IIb: Argument of Triviality—Spatiotemporal Features versus Spatiotemporal Mechanisms

I argue that the proponent of the argument of triviality neglects the difference between spatiotemporal features on one hand and spatiotemporal mechanisms on the other. Yes, it is true that the brain exhibits spatiotemporal features including different frequency ranges and regions as we can observe them. Therefore, to now characterize the brain by these spatiotemporal features is correct, but it does not say anything beyond what is evident; the spatiotemporal characterization of the brain is thus trivial. The argument of triviality may thus hold for the brain’s spatiotemporal features. However, even if this were the case, the argument of triviality for the brain’s spatiotemporal features does not imply that the same holds for spatiotemporal mechanisms and thus for consciousness itself.

We may now turn our consideration to the exact nature of the relation between spatiotemporal features and spatiotemporal mechanisms.

How are spatiotemporal features and mechanisms related to each other? The data show that the same spatiotemporal features, such as the same regions and the same frequency range, can be related to two different spatiotemporal mechanisms, such as spatiotemporal integration and fragmentation, with different outcomes or results, exemplified by spatiotemporal nestedness and isolation. I propose, then, that spatiotemporal features and mechanisms can dissociate from each other: the same spatiotemporal features can be associated with different spatiotemporal mechanisms in the same way that different spatiotemporal features can be related to one and the same spatiotemporal mechanism.

What does such possible dissociation between spatiotemporal features and mechanisms indicate for the argument of triviality? The argument of triviality certainly applies to spatiotemporal features. It is indeed trivial to characterize the brain by different regions or different frequencies, since those spatiotemporal features determine the brain as brain. The absence of those spatiotemporal features portends the absence of the brain, that is, brain death. Hence, a proponent of the argument of triviality is correct in stating the trivial nature of spatiotemporal features.

However, this does not imply that the proponent of triviality is also correct when it comes to an analysis of the genesis of spatiotemporal mechanisms. Spatiotemporal mechanisms are not identical with spatiotemporal features. This is well reflected in the fact that the same spatiotemporal features, such as different frequencies, can be associated with different spatiotemporal mechanisms, such as spatiotemporal nestedness or isolation (as demonstrated in the data in the first part of this chapter). We therefore need to describe and specify the spatiotemporal mechanisms in order to understand how, as a result, one and the same set of spatiotemporal features can lead ultimately to different results or outcomes, for example either to spatiotemporal nestedness or isolation.

Given that spatiotemporal features and spatiotemporal mechanisms are not identical, the characterization of the brain’s neural activity by spatiotemporal mechanisms adds something novel and is thus no longer to be considered trivial and self-evident. The argument of triviality can consequently be rejected with regard to spatiotemporal mechanisms (while, at the same time, we may accept triviality for spatiotemporal features). Thus, when the proponent of this argument declares the spatiotemporal model of consciousness to be trivial, he or she is confusing spatiotemporal features and mechanisms in the brain’s neural activity.

Importantly, the nontrivial characterization of the brain by spatiotemporal mechanisms carries over to consciousness. The data discussed throughout this chapter characterize consciousness by specific spatiotemporal mechanisms: spatiotemporal nestedness, expansion, and globalization. I have showed that deficits in these spatiotemporal mechanisms lead to the absence of consciousness even if the spatiotemporal features of the brain remain the same. Spatiotemporal mechanisms are thus critically relevant for the presence of consciousness. Since spatiotemporal mechanisms are critically relevant, the spatiotemporal characterization of consciousness is not trivial at all.

Spatiotemporal Model IIc: Argument of Triviality—Phenomenal and Ontological Implications

The proponent of the argument of triviality may not yet give in. Even if nontrivial on empirical and conceptual–logical grounds, the spatiotemporal characterization of consciousness nevertheless may appear trivial when it comes to phenomenal and ontological issues. The spatiotemporal characterization simply does not add anything and only accentuates the triviality: consciousness occurs in the space and time of the world and must therefore be characterized as spatiotemporal by default. I, obviously, for the reasons explained in the last three paragraphs, reject that argument.

Regarding phenomenal features, the spatiotemporal model of consciousness is far from trivial. The spatiotemporal model of consciousness carries the implication that phenomenal features such as qualia and intentionality, among others are indeed spatiotemporal rather than nontemporal and nonspatial as is often assumed (explicitly or, more often, implicitly) in philosophy. If the hypothesis of spatiotemporal correspondence between neuronal and phenomenal features is correct, the phenomenal features may, for instance, be characterized by spatiotemporal nestedness of the various contents in consciousness as well as by spatiotemporal expansion of specific contents. In contrast, isolated contents that are not spatiotemporally nested within and expanded by the brain’s spatiotemporal structure will then remain impossible in consciousness.

Accordingly, the spatiotemporal model of consciousness leads to the neurophenomenal hypothesis (Northoff, 2014b), which can be experimentally tested to make specific predictions about the relation between neuronal and phenomenal features (see Northoff, 2014b, for full development of this thesis). The spatiotemporal characterization of consciousness is thus far from trivial when it comes to its phenomenal features.

How may we categorize ontological issues? Time and space are presupposed and defined in a certain way pertaining to how the brain’s neural activity itself constructs time and space. We saw that the brain relates and links different temporal and spatial scales with each other in a way that results in spatiotemporal integration and nestedness (see first part in this chapter). Such spatiotemporal integration and nestedness presuppose a particular ontological notion of time and space: time and space can no longer be accounted for in terms of discrete points in time and space but, rather, by their relation and structure. This ontologic confluence is what I describe as relational time and space (discussed in greater detail in chapter 9) that presupposes relation and structure rather than elements and properties as the basic units of existence and reality. In essence, this presents as a theory of ontic structural realism, which is the overarching subject of chapters 9–11. Accordingly, the spatiotemporal model of consciousness is far from trivial when it comes to considering ontological matters and presenting major ontological implications for explication of the mind–body problem.

Ultimately, then, I reject the argument of triviality on empirical, conceptual-logical, phenomenal, and ontological grounds. Empirically, spatiotemporal mechanisms are not trivial since they must be distinguished from other mechanisms such as cognitive, sensory, and other mechanisms. Conceptually and logically, we must distinguish spatiotemporal mechanisms from spatiotemporal features with only the latter but not the former being considered as trivial. While taken in the context of experience, that is, in a phenomenological context, the spatiotemporal model implies spatiotemporal characterization of phenomenal features—a process that absolutely cannot be characterized as trivial given that phenomenal features are often determined to be nontemporal and nonspatial. Finally, the spatiotemporal model cannot be considered trivial on ontological grounds, since it presupposes different nontraditional determinations of both time/space and existence and of reality in general.

Conclusion

Time and space are the central and most basic building blocks of nature, and therefore they naturally apply to the brain and how the brain constitutes its neural activity. For this reason, we here propose a conception of the brain’s neural activity in primarily spatiotemporal terms that we hypothesize are central for consciousness and that thereby constitute our spatiotemporal theory of consciousness.

How, it is essential for us to ask, can stimuli and their respective contents be associated with consciousness? Consciousness goes beyond contents as we have discussed in chapters 5 and 6. Yet what exactly does this “beyond” consist in? In this chapter I have argued that this “beyond” consists of spatiotemporal features: the spatiotemporal mechanisms of the brain’s neural activity that allow neural matter to associate stimuli with consciousness. In this chapter we have discussed three such spatiotemporal mechanisms, spatiotemporal nestedness, spatiotemporal expansion, and spatiotemporal globalization. Despite their differences, all mechanisms are intrinsically spatiotemporal, that is, they are spatiotemporal mechanisms, as distinguished from content-based, cognitive, and informational mechanisms. Spatiotemporal mechanisms allow for neural activity and stimuli to be processed according to their spatiotemporal features: they allow stimuli and their respective contents to be processed in a more expanded and nested spatiotemporal context that allows the association of these stimuli with consciousness.

More generally, spatiotemporal expansion and nestedness can be considered as the empirical building blocks of a spatiotemporal model of consciousness (see Northoff & Huang, in press, as well as Northoff, 2017a,b, for more empirical details in what we describe as the spatiotemporal theory of consciousness or STC). In essence, the spatiotemporal theory of consciousness conceives both brain and consciousness in spatiotemporal terms rather than in terms of sensorimotor, cognitive, affective, or social functions and their respective contents. I have demonstrated that such a spatiotemporal model of consciousness can well counter the challenges of the arguments of both nonspecificity and triviality. Most important, as proposed in the concluding section of this chapter, the spatiotemporal model of consciousness is not only empirically relevant but also carries major conceptual, phenomenological, and ontological implications—and these topics become the focus for the next part of this book.

Acknowledgment

I thank Zirui Huang with whom I recently published several parts of this chapter (Northoff & Huang, 2017, in Neuroscience & Biobehavioral Reviews).