While avoiding philosophical definitions of consciousness (e.g., James, 1890/1950; Searle, 1994), which are beyond the scope of this paper, the term consciousness here generally denotes, as in previous neuroscientific approaches (Edelman, 2006; Boly et al., 2009), an experienced property of mental states and processes, which is lost during a dreamless deep sleep, deep anesthesia or coma. Consciousness and self-consciousness are tightly related, based on both philosophical accounts and cognitive theories (e.g., Gennaro, 1996; Natsoulas, 1998; Kriegel, 2004; Morin, 2006; Gallagher and Zahavi, 2008; Damasio, 2012). We are aware that the concept of self has many definitions and that there is no consensual framework for conceptualizing the various aspects of the self. Yet, we adopt here an increasingly accepted framework for the self, grounded in James' (1890/1950) differentiation between the self as “I,” the subjective knower, a momentary enduring presence, and the self as “me,” the object that is known, the self-concept and autobiographical identity. This framework distinguishes between the “minimal self” (MS), a self that is short of temporal extension, which is endowed with a sense of agency, ownership, and non-conceptual first-person content, and the “narrative self” (NS), which involves personal identity and continuity across time, as well as conceptual thought (Gallagher, 2000). Consciousness can also be divided into a simpler and a more complex form, each one of them supporting one type of self-experience. The first is core-consciousness (CC), which supports the MS, its scope being the here and now. The second is extended-consciousness (EC), which supports the NS, and involves memory of past, imagination of future, and verbal thought (Damasio, 1999, 2012). Importantly, while CC is independent of the EC, and relies only on its exchange with the body (and environment), the EC is always dependent on CC (Damasio, 1999). Hence, the NS is dependent on the MS, but not vice versa.

In cognitive neuroscience the MS and NS have been attributed to various different neural processes. Following, we will refer to these neural spaces as N_ms and N_ns, respectively. While these neural spaces cannot yet be fully identified, there is accumulated knowledge suggesting main brain regions involved. NS has been conceptualized as self-referential processing, such as assessing one's personality, appearance or feelings and recognizing one's own face or name. The neural regions supporting self-referential processing are mainly the midline regions, including the medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC) and precuneus (Gusnard et al., 2001; Northoff and Bermpohl, 2004; Northoff et al., 2006), as well as the temporoparietal junction (TPJ) and temporal pole (Christoff et al., 2011). The MS has been attributed to self-specifying processing, experiencing oneself as the agent of perception, action, cognition and emotion. The cortical regions suggested to be involved include those related to sensorimotor integration (such as motor and supplementary motor area—SMA) and proprioception (the insula), as well as higher-level regulatory regions, including dorsal anterior cingulate cortex (dACC) and dorsolateral PFC (DLPFC) (Legrand, 2006; Legrand and Ruby, 2009; Christoff et al., 2011). Other regions involved in the sense of agency include the TPJ and inferior parietal lobule (IPL) (Chaminade and Decety, 2002; Blanke and Metzinger, 2009).

Intriguingly, N_ms and N_ns can be related to a dual organization of the cortex. Accumulating evidence supports this notion, showing that thalamo-cortical networks can be divided into two, often antagonistic, global systems (Fox et al., 2005; Golland et al., 2007; Tian et al., 2007; Soddu et al., 2009): (i) a system of inward-oriented networks (the “intrinsic” or default mode network - DMN); and (ii) a system of externally-oriented, sensory-motor networks (the “extrinsic” system). Resting-state activity involves the DMN (Raichle et al., 2001; Greicius et al., 2003), a task-inhibited network related to self-reference and mind-wandering. The DMN includes a consistent set of five regions that comprise the mPFC, PCC, IPL, medial temporal lobe (MTL) including the hippocampus, and lateral temporal cortex (LTC) (Buckner et al., 2008). Task-induced neural activity is related to the dorsal attention network (DAN), which includes regions in the frontal eye fields, ventral premotor cortex, the supplementary motor area (SMA), superior parietal lobule, intraparietal sulcus, and motion-sensitive middle temporal area (Corbetta et al., 2008). Interposed between them is suggested to be the frontoparietal network (FPN), which includes the anterior PFC, DLPFC, dorsomedial superior frontal, ACC, anterior IPL, and anterior insular cortex (Vincent et al., 2008). The FPN cooperates with either one of these typically antagonistic systems, possibly integrating information from, and adjudicating between, these two potentially competing brain systems (Vincent et al., 2008; Spreng et al., 2010; Smallwood et al., 2012). The FPN can be broken down into two sub-networks (Seeley et al., 2007), the “executive control network” (DLPFC and IPL) and the “salience system” (anterior insula and ACC), with the latter also being specifically attributed the role of switching between the intrinsic and extrinsic systems (Menon and Uddin, 2010). Following, we will refer to this interposed network generally as N_i.

Based on the dual organization of consciousness, self, and underlying neural activity, CSS is organized into two concentric spheres around the body. The concentric organization depicts the reliance of each sphere on the previous level: CC relies on the body, while EC relies on CC. The inner sphere of CC/MS is phenomenologically related to the body, experiencing agency and momentary sensory experiences. It is embodied. The inner sphere is surrounded by the EC/NS sphere, which is phenomenologically further away from the body, in the realm of conceptual thought, language, memories and imagination, and relies more on mental representation, rather than actual sensory experiencing (Figure 1A).

We refer more generally to the neural space, mainly building on current neuroscientific knowledge. As to the neural space of the CC/MS (N_ms), this is the point where the physiological condition of the body, which is in constant exchange with the environment, becomes available to the brain. This mainly relies on sensori-motor integration, i.e., convergence of action and perception, allowing one to perceive the sensory consequences of one's action through action monitoring, and proprioception—perceiving the body state. The proposed regions of the brain involved in sensori-motor integration are SMA and pre SMA (Legrand, 2006; Ferri et al., 2012), while proprioception involves the somatosensory and insular cortex (Craig, 2002, 2009) as well as a deep portion of the posterior cingulate (Parvizi et al., 2006; Damasio and Meyer, 2009). The neural reference space of the EC/NS (N_ns) is largely suggested to involve the DMN. It should be noted, however, that some posterior regions of the DMN, including the IPL and precuneus, are argued to be involved in both NS and MS due to their roles in agency (Chaminade and Decety, 2002) and CC (Damasio and Meyer, 2009), respectively. These regions can be viewed as a mutual reference space for both spheres. Between the N_ms and N_ns there is the interposed Ni, largely related to the control FPN system, which shifts between collaboration with both (Figure 1B).

Another important feature of the CSS is the existence of two simultaneous trajectories of experience, each in one of the spheres. This will be further developed in sections Second Dimension of the CSS—Awareness and The Dynamics within the CSS.

Consciousness would be inconceivable without temporality, as time is an omni-present structural feature of consciousness (James, 1890/1950). As James wrote: “The knowledge of some other part of the stream [of consciousness], past or future, near or remote, is always mixed in with our knowledge of the present thing…. These lingerings of old objects, these incomings of new, are the germs of memory and expectation, the retrospective and the prospective sense of time. They give that continuity to consciousness” (p. 606–607). Past or future events can be activated in experience voluntarily, and this constant mental time travel aids one in understanding the meaning of present happenings: “I don't simply exist in the present and happen to have the capacity to envisage the future and remember the past. Rather, human reality is characterized by a kind of temporal stretch. The past continually serves as the horizon and background of our present experience, and when absorbed in action, our focus, the center of our concern, is not on the present, but on the future goals that we intend or project” (Gallagher and Zahavi, 2008, p. 86).

We claim that this time dimension is sufficient, and can also account for the spatial aspect of experience. First, at any time point, one sees the world from only one spatial perspective (Revonsuo, 2003). Second, each episodic memory has one (and only one) spatiotemporal content (Russell and Davies, 2012). Third, the phenomenal fields of different modalities are spatially and temporally integrated, so that different features belonging to the same object are realized in the same location and time (Fingelkurts et al., 2010). Thus, spatial experience is fully integrated with temporal experience in three important ways.

Several cognitive models have been proposed for time-consciousness, most of them variants of a pacemaker–accumulator clock, where experienced duration is represented by a pacemaker which produces a series of pulses recorded over a given time span (Zakay and Block, 1997; Glicksohn, 2001). Yet, competing cognitive models propose that memory decay processes are involved in time perception (Wackermann and Ehm, 2006). A contrasting view to the cognitive models is Varela's (1999) dynamic model of the experience of time, according to which three different scales of duration contribute to a cognitive act: the elementary (10–100 ms), integration (0.5–3 s) and narrative (>10 s) scales. Neurophysiologically, the first two correspond to neuron-level electrophysiology and synchronization. These further correspond to the experienced present. Here, we adopt a dynamic view, akin to Varela's view. However, we differentiate between two time scales, combining Varela's elementary and integration scales into one: the immediate perception of the present moment (<3 s; see Pöppel, 1997), contrasted with the longer time scale. The longer time scale refers to the re-presentation of experience, while the second refers to the immediate perception of the present moment (Figure 3A). By “re-presentation” we mean that experience, with its full-blown, present-moment, multi-dimensional vividness, is being “projected,” or re-presented (and not represented) into another subjective time, either past or future. The fact that we usually cannot both recall and be “here and now” simultaneously is manifested by placing each of the two phenomenological categories into the two different spheres of CSS. Following, we describe in more detail these two categories along the time continuum.

The first temporal category refers to the longer time scale, and involves the re-presentation of experience in the past and in the future. Unlike immediate perception, this is psychologically further away from the body: when one's conscious awareness re-lives the past or the future, one's conscious awareness is decoupled from the body (which experiences the now). It encompasses mental re-presentations of other “nows,” relived or imagined. The intriguing ability of the human mind to mentally travel through time, enabling one to relive past experiences through memory, or project oneself into the future by generating a prediction based on memory, has been also referred to as autonoetic (self-knowing) consciousness (Wheeler et al., 1997; Stuss et al., 2001; Markowitsch, 2003). Hence, this experience pertains to the NS, and is within the EC/NS sphere. At the past end, we find re-presentation of the far or near past, by means of retrospective memory retrieval. This includes either true or false memories, which are experienced as being phenomenologically similar (Lampinen et al., 1997). Here, we adopt Conway's phenomenological description of autobiographic memory (Conway and Pleydell-Pearce, 2000; Conway, 2009), as well as Morin's (2006) conceptualization of long-term memory. Conway relates autobiographic memory with the NS (Conway, 2005). Autobiographic memory consists of recollected episodes from one's life, which are based on both episodic as well as semantic memory (Tulving, 1985; Conway and Pleydell-Pearce, 2000; Conway, 2005). Autobiographic knowledge comprises several levels of categorization. First is event-specific knowledge, which is a summary record of sensory-perceptual–conceptual-affective processing derived from working memory, which is predominately re-presented in the form of visual images (for example, a specific restaurant). Event-specific knowledge, in turn, is contextualized within a general event (e.g., during the vacation in Greece). The general event, in turn, is associated with one or more lifetime periods that locate the more specific knowledge within an individual's autobiographical memory as a whole (e.g., it was just after our marriage) (Conway and Pleydell-Pearce, 2000; Conway, 2005). What is common to all these levels of autobiographic memory is the phenomenal feeling of remembering: “the feeling signals the state in an experiential way. Recollective experience, the sense of the self in the past and the episodic imagery that accompanies that sense, indicate to the rememberer that they are in fact remembering and not daydreaming, fantasying, or in some other non-memory state” (Conway, 2005, p. 614). At the other end of the time continuum, the future, we find prospective memory, which is memory for future intentions (Glicksohn and Myslobodsky, 2006), and refers to the functions that enables a person to carry out an intended act after a delay (Burgess et al., 2001). Another ubiquitous phenomenon is the generation of predictions and future simulations based on previous autobiographic knowledge (Schacter and Addis, 2007; Schacter et al., 2007), sometimes termed “proaction” (Bar, 2007, 2009). Mental time-travelling to the future and the past can be actually “experienced” in the “here and now” (Gilbert and Wilson, 2007): if the mental simulation is strong enough, the imagined or recollected images can evoke bodily reactions. At the same time, actual sensory stimulation from the environment is blocked, and one experiences sensory decoupling from the environment (Smallwood et al., 2007).

We now turn to the second category, the immediate perception of the present moment (<3 s), which pertains to the CC/MS sphere. Various models for the phenomenology of immediate time perception have been proposed (Dainton, 2008). One major category is retentional models: our experiencing of change occurs within episodes of consciousness which themselves lack temporal extension, but whose contents coordinate with past and future by virtue of their place in the temporal structure (Dainton, 2008). Specifically, we adopt the retentionalist model for consciousness of time outlined by Gallagher and Zahavi (2008), which assumes a Husserlian view: the immediate sensation, or the “primal impression” is combined with retention (being aware of the “just-passed” slice of the experience) and protention (being aware of the “just-about-to-be”). A perception cannot merely be a perception of what is now, but must include a retention of the just-passed and a protention of what is about to occur. Importantly, retention and protention are not memory, or imagination, which re-present the experience. Rather, they are actual experience. Unlike long-term memory and expectation, they are involuntary and automatic processes, and they could be argued to be working memory (Vogeley and Kupke, 2007; Gallagher and Zahavi, 2008). This experience is related in our model to the MS.

Turning to the neural space, the first category of autonoetic consciousness, within the EC/NS sphere, should involve the N_ns. In contrast, CSS predicts that the immediate perception of the present, within the CC/MS sphere, is related to the N_ms and bodily processing. As subsequently presented, these predictions are confirmed by neuroscientific evidence.

The most important neural structure for memory is the hippocampus, the locus of interaction between working memory and long-term memory (Fell and Axmacher, 2011). It has been proposed that memory initially depends on the hippocampus. However, with increasing time, the hippocampus becomes less important, and the involvement of multiple cortical regions increases, including the medial frontal gyrus and precuneus (Smith and Squire, 2009). Yet, new findings confirm the important role of the hippocampus even in retrieval of long-term, established memories, in collaboration with the ACC (Suzuki and Naya, 2011). In a nutshell, the hippocampal complex is essential for encoding, retaining, and recovering experiences, enabling the immediate subjective and vivid experience. Other regions, mainly the prefrontal cortex, select, organize, help retrieve, monitor, and verify the hippocampal recollection (Moscovitch, 2008). Though both recent and remote memories are associated with hippocampal activation, it was found that activations associated with more recent memories cluster at the anterior hippocampus, whereas those associated with more remote memories are distributed across its length (Gilboa et al., 2004). Not only memory, but also planning involves the hippocampus, as well as frontal and parietal structures. Strikingly, there is an overlap between memory systems and the network involved in foresight, and these two overlapping regions also overlap with the DMN (key component in N_ns), including the hippocampus, mPFC, precuneus and lateral parietal cortex (Bar, 2007, 2010; Schacter and Addis, 2007; reviewed by Schacter et al., 2007). Another line of research, on mental time traveling and “self-projection,” revealed the involvement of the IPL (Nyberg et al., 2010), and the temporo-parietal junction (Arzy et al., 2009), which are key regions for self-referential processing and which are considered components of the DMN.

Turning now to the cortical regions, which have been suggested to be involved in the immediate perception of the present moment, Rubia and Smith (2004) emphasize the DLPFC, ACC, SMA, and IPL in their review of the literature. The IPL is also strongly suggested by others who review the literature (Walsh, 2003; Oliveri et al., 2009). Another suggested region is the insula (Craig, 2002, 2009; Wittmann, 2009), relating cognition of duration with proprioception. This proposition was supported by a functional magnetic resonance imaging (fMRI) study, showing a linear build-up of neuronal activation in the insula during a time reproduction task (Wittmann et al., 2010) and by an anatomical study (Gilaie-Dotan et al., 2011), showing that the gray matter volume of the right sensory cortex is correlated with the ability to discriminate time intervals. In addition to the literature on time perception, we consider Baddeley's (1992, 2003) influential model of working memory. Here, again, we find that the DLPFC plays an important role, as an executive control system, assisted by two subsidiary storage systems: the phonological loop and the visuospatial sketchpad (including right and left IPL and premotor cortex, respectively), both of which store perceptual information. Indeed, the DLPFC is believed to provide a buffer to hold information in mind, and to order it in space-time (Dehaene and Naccache, 2001). To conclude, as hypothesized, all the neural regions that are related to momentary experience of time, as well as to working memory, are within the FPN and DAN, considered as key elements in N_ms and N_i, and in contrast, autonoetic consciousness and prospective memory involves the DMN, as a key element in N_ns.

Awareness is a primary feature of consciousness, being the subjective experience of internal phenomena, a perception of the field of inner and outer events that encompasses one's reality at any given moment, the state of perceiving (Laureys, 2005; Cohen and Dennett, 2011). Awareness can be largely categorized into two types, following the influential conceptualization of Block (1995, 2007), represented in CSS as the two sides of the awareness dimension. The first is access awareness, which corresponds to states that can be reported on, by virtue of high-level cognitive functions such as memory and decision making, and which necessitates attention. The second is phenomenal awareness, related to private first-person experience, and occurs without—or with very little—attention (Kouider et al., 2010). In contrast to this division, awareness could also be conceived of as a graded phenomenon (Kouider et al., 2010): at one end expanded awareness, when all levels of relevant processing are accessible, and at the other end complete non-awareness, when all levels of processing are not accessible. Intermediately, there is partial awareness, combining awareness at some level and unawareness at another level of processing.

A possible solution to the current debate whether top-down attention is necessary for consciousness (Cohen et al., 2012a,b) or if these are two independent processes (Koch and Tsuchiya, 2007a,b; Tsuchiya et al., 2012) could be settled if we consider that phenomenal awareness can emerge without top-down attention (Aru and Bachmann, 2013), in contrast to access awareness. This would support the notion that attention allows information to be more fully transmitted across cortical regions than unattended information (Cohen et al., 2012b), hence is required for access awareness. This argument supports our proposition that the awareness dimension stands for attention as well.

In the following, we describe various phenomenal states along the awareness continuum, moving from no phenomenal access to high phenomenal access (Figure 3B). In doing so, we largely rely on Gallagher and Zahavi's (2008) account of pre-reflective and reflective consciousness, as well as on Morin's (2006) social/personality model, describing degrees of consciousness based on several theories (including Brown, 1977; Natsoulas, 1998; Schooler, 2002). Starting at no phenomenal access, we find outside CSS states of consciousness in which there is no phenomenal awareness to either external or internal input. These include the dreamless portion of deep sleep, coma, anesthesia, vegetative state, epileptic loss of consciousness, and somnambulism. When one regains awareness, there is still no access or memory as to external or internal happenings during that state. Further along the continuum, there are preconscious states and subliminal experiences. A division should be drawn between inaccessible internal and external input. Processing of internal input is conducted within the EC/NS sphere, and is referred to here as “subliminal awareness.” This includes normal states of day dreaming, as well as pathological states such as dissociation. Processing of external input is conducted within the CC/MS sphere, and is referred to as “subliminal sensory awareness.” These states include the natural decoupling of attention from sensory processing (Smallwood et al., 2007), as can occur during driving. All of these states are not accompanied by top-down attention.

Next, we describe the access states along the awareness axis, which can be conceptual or non-conceptual (Kapitan, 2006), as subsequently detailed. First, within the CC/MS sphere, there is first-order awareness, also called pre-reflective awareness (Gallagher and Zahavi, 2008). This is an implicit and direct awareness to experience, prior to any reflection on the experience. In this state, according to Morin (2006), one will directly be attentive and process external input from the environment, without conceptual elaboration of the mental events that are taking place. Hence, the organism will be totally immersed in experience. These states are accompanied by top-down attention to external input (Chun et al., 2011). States along the access awareness within the EC/NS sphere involve second-order awareness, also called reflective awareness (Gallagher and Zahavi, 2008). This is an explicit, conceptual, and objectifying awareness, which is accompanied by focal attention to internally generated input (Chun et al., 2011). In this state, one attends directly to the cognitive experience itself. It is described by Morin (2006) as the capacity to become the object of one's own attention, a process that occurs when an organism focuses not on the external environment, but on the internal milieu. In its extreme form, it becomes meta-awareness, being aware that one is aware (Morin, 2006).

We rely on neuronal global workspace theory (Dehaene and Naccache, 2001; Dehaene et al., 2006) to describe the corresponding neural space, suggesting that conscious access is produced through the interaction between specialized neural subsystems and a multimodal limited capacity global workspace (Baars, 2002, 2005), the FPN (the key component of N_i). During states of phenomenal awareness, including subliminal and preconscious states, activation due to internal or external input does not involve the FPN. In contrast, during states of access awareness, synchronized activity increases in the FPN, which becomes capable of guiding intentional actions including the production of verbal reports. The transition between phenomenal and access awareness is sharp, as expected in non-linear dynamic systems (Dehaene et al., 2006). Furthermore, CSS posits two simultaneous trajectories, one in the EC and the other in the CC. As these two trajectories are usually antagonistic, habitually one has access awareness only to the phenomenology related with one of the trajectories, while the phenomenology related with the second trajectory continues its activity at a sub-threshold level, within the phenomenal awareness continuum (Figure 4). Hence, the brain alternates dynamically, shifting awareness from internal to external processing. The actual phenomenal experience is constantly dictated by the neural space that is more active at the moment, and which is synchronized with the N_i. When one is immersed in “intrinsic” (i.e., N_ns) activity, one is “decoupled” from extrinsic processing (i.e., N_ms), which nevertheless continues (Smallwood et al., 2007; Vanhaudenhuyse et al., 2011). And vice versa, while externally engaged, the intrinsic system is continuously activated in a “sub-threshold for awareness” manner. Indeed this is what Singer has been proposing from the early 60s, in his exposition of daydreaming (Singer, 1966). As he has more recently suggested (Singer, 2009, p. 196), “Many years ago I proposed that what we now call the brain's default network may be almost continuously active at a subthreshold level.” The FPN alternates between cooperation with the intrinsic or the extrinsic systems. This notion is supported by the findings that a systematic impairment of FPN was found in altered states of consciousness, such as sleep, anesthesia, coma, vegetative state, epileptic loss of consciousness, and somnambulism (summarized by Boly et al., 2009), all of these being states which are considered outside the model's awareness dimension. This suggests that an intact FPN, enabling access awareness, is a prerequisite for a normally functioning CSS.

In relation to this neural space, both second-order awareness (within the EC/NS sphere, upper side of continuum) and mind wandering, i.e., subliminal awareness (within the EC/NS sphere, lower side of continuum) have been related to DMN activity (N_ns). However, what differentiates between states has been shown to be DLPFC activation, which characterizes thoughts that occur with access awareness (Smith et al., 2006; Christoff et al., 2009). The suggested role of FPN as switching collaboration between the two neural spaces which support the two phenomenological spheres along the access awareness is illustrated by the following studies. In relation to the inner sphere of CC/MS, an fMRI study (Ferri et al., 2012) showed that the “bodily-self” (self rooted in bodily motor experience) recruits pre SMA, SMA and insular regions, belonging both to the FPN and DAN systems (N_i and N_ms, respectively). In relation to the outer sphere of EC/NS, an fMRI study emphasized the importance of the DLPFC (N_i), as participants learned to regulate its activation by turning their attention toward and away from the contents of their own thoughts, or their DMN-intrinsic (N_ns) system (McCaig et al., 2011).

Emotion and consciousness are considered by many to be inseparable (e.g., Damasio, 1999; Lambie and Marcel, 2002; Panksepp, 2005; Barrett et al., 2007; Tsuchiya and Adolphs, 2007), as each conscious state is endowed with some form of emotion, to the point that even the perceptual representation of everyday objects carries subtle affective tone (Lebrecht et al., 2012). Put in the words of Searle (1994, p. 7), “part of every normal conscious experience is the mood that pervades the experience. It need not be a mood that has a particular name to it, like depression or elation; but there is always what one might call a flavor or tone to any normal set of conscious states.”

Emotion states are generally agreed to bear two important phenomenal features, the one is mental and the other bodily, grossly speaking. Hedonicity is both intrinsic to bodily states, and depends on the interpretation placed on them. A similar differentiation has been done with many, albeit calling it by various names. Schachter and Singer (1962) described emotional experience as a combination of general arousal, and the cognitive attribution of the cause of this arousal. Similarly, Mandler (1984) distinguished between non-specific arousal, awareness of which provides the intensity of the emotional experience, and the evaluative structure (cognitive interpretation of the situation), which provides the particular content and quality of emotional experience. Damasio (1999) views emotional experience as consisting of sensory changes that occur in the viscera and internal milieu (which he calls emotions) and the mental image of these sensory patterns (which he calls feelings). According to Lambie and Marcel (2002), any emotional state is defined by a combination of two things: the bodily action readiness (and its representation) and the evaluative description (a mental representation). The first includes certain bodily and brain systems which are activated in response to stimuli (chiefly the limbic, autonomic, hormonal and aspects of the skeletal nervous system). The second, the evaluative description, is an appraisal leaving a record, a description of how one's concerns or one's self have been affected by the event. This is a mental representation. Barrett's (Barrett, 2006; Barrett et al., 2007) view of emotion involves two operations. The first is called “core-affect,” which includes bodily fluctuations that are represented in the brain. The second is “conceptualization,” a process by which stored representations of prior experiences (i.e., memories, knowledge) are used to make meaning out of sensations in the moment. To summarize, there is general agreement that emotional states bear two important phenomenal features, one mental and the other bodily. Following, we will refer to these two components of emotion as valence (a subjective feeling of pleasantness or unpleasantness) and arousal (extent of bodily excitation), respectively.

We suggest that the manifestation in CSS of these two phenomenological qualities of the emotional experience, namely arousal and valence, is within the CC/MS and EC/NS spheres, respectively (Figure 3C). Within the CC sphere, arousal increases in two directions stemming from a minimal degree of arousal at the center of the emotion continuum. Valence, manifested within the EC sphere, increases toward the pleasant and unpleasant ends of the continuum. Each emotional experience has an arousal component, namely a bodily and sensory element, and valence, or a mental representation, in line with the above formulations of the phenomenology of emotions, and with the two CSS trajectories. The relationship between valence and arousal has been subject to a long debate, and various models have been proposed. A classical description is Russell's (1980) circumplex, suggesting an orthogonal relationship. In another framework, Lewis et al. (2007) suggest that aspects of valence generate a U-shaped curve with arousal. According to such a framework, strong positive valence is accompanied by strong positive arousal, and similarly for the negative aspects. Yet, others suggest that the dissociation between valence and arousal might be an issue of measurement more than reflecting distinct qualia underlying emotional experience (Larsen et al., 2003; Kron et al., 2013). For example, Russell (1989) has explicitly opposed the idea that arousal is only a physiological concept, writing (p. 106), “… there is no more reason to speak of arousal as strictly physiological and pleasure-displeasure as strictly mental then there is to express it the other way around.” In comparison to the ongoing debate, the CSS model accounts for both valence and arousal, and allows for an orthogonal relationship between them. This means that each point on the arousal continuum could be in principle accompanied by any simultaneous point on the valence continuum, as it depends on the interpretation placed on it, which might be multiple for any given bodily state (as accounted by Schachter and Singer, 1962). As for the Lewis et al. U-shaped proposed relationship, we propose this should be rejected because negative arousal can sometimes be accompanied by positive mental evaluation, as in Schachter and Singer (1962), or as in the example of watching safely a horror movie. Further, the interplay between the emotion and awareness dimensions within CSS predicts a novel relationship between arousal and valence, namely antagonism: emotional states involve both evaluation and the products of arousal, but both of these need not be simultaneously present in experience or awareness, and certainly are not always experienced as such. This means that one can have access awareness to either the arousal or valence aspect at each specific moment and state, and that access awareness alternates between them. As a trait, one can be more prone to emphasize either the arousal or the valence, as shown by Barrett (1998). In addition, the interplay between the emotion and awareness dimensions creates a wide spectrum wherein emotional experience can be classified in terms of accessibility, as suggested by Lambie and Marcel (2002), Damasio (1999) and Frijda (2009), from non-conscious emotions, through phenomenal emotions (1st-order emotional experience), to awareness of emotional experience (2nd-order, emotional experience).

CSS predicts that the neural space for the arousal component of emotion should be related to the N_i and N_ms (their key components being DAN and FPN, respectively), while the valence component should be related to the N_ns (i.e., DMN). The evidence for that is based on a meta-analysis conducted by Barrett and colleagues (Barrett et al., 2007; Kober et al., 2008; Lindquist et al., 2012), stemming from a constructionist approach to emotion, where the assumption is that emotional mental states result from an interplay of more basic psychological processes that may not, themselves, be specific to emotion (Barrett et al., 2007; Kober et al., 2008; Lindquist et al., 2012). According to this meta-analysis, the distributed network involved in realizing core-affect includes several sub-cortical, as well as cortical regions: the amygdala, which signals whether exteroceptive sensory information is motivationally salient; the anterior insula, which plays a key role in representing core affective feelings in awareness based on its role in the awareness of bodily sensations and affective feelings; portions of the orbitofrontal cortex, as a site that integrates exteroceptive and interoceptive sensory information to guide behavior; ACC, and more specifically subgenual ACC, regulating somatovisceral states, pregenual ACC, as a visceromotor (i.e., autonomic) control area involved in resolving which sensory input influences the body when there are multiple sources of sensory input, and anterior midcingulate cortex (a part of the FPN), delivering sources of exteroceptive and interoceptive sensory information to direct attention and motor response. The regions suggested by this constructionist approach fall largely within the interoceptive/exteroceptive processing network, as well as the FPN, as suggested by CSS.

In Barrett's (Barrett et al., 2007; Lindquist et al., 2012) view, the mental representation (valence) is specifically related to the DMN: “In our model, categorization in the form of situated conceptualization is realized in a set of brain regions that reconstitutes prior experiences for use in the present. This set of brain regions has also been called the ‘episodic memory network’ or the ‘default network’… this psychological operation makes a prediction about what caused core affective changes within one's own body or what caused the affective cues (e.g., facial actions, body postures, or vocal acoustics) in another person, and this prediction occurs in a context-sensitive way (with the result that core affect in context is categorized as an instance of anger, disgust, or fear” (Lindquist et al., 2012, p. 129). Another important network for categorization and emotional perception includes the anterior temporal lobe (ATL) and ventrolateral prefrontal cortex (VLPFC). Further, the DLPFC is postulated as being involved in mental states of attending to emotional feelings or perceptions, and holding affective information in mind in order to categorize it (Lindquist et al., 2012). Hence, the constructionist approach to emotion provides support for CSS, both regarding the categorization of emotions as being coupled with the DMN, as well as attributing to the FPN the role of mediating the activity within the two CSS spheres.

The CSS model is a dynamic system, with rich self-organizing properties. One novel aspect of CSS is the suggestion that each sphere functions as a separate dynamic system, with its own trajectory over time (Figure 4A). The two trajectories are simultaneously present, one within the inner sphere of CC/MS, and the other in the surrounding sphere of the EC/NS. These trajectories are usually antagonistic, and phenomenal awareness switches between them, as elaborated in section Second Dimension of the CSS—Awareness. In the neural space, this is manifested by the collaboration, through synchronization, of the N_i with either N_ms or N_ns (Figure 4B). This antagonistic behavior, however, should not be seen in early childhood.

While the detailed ontogenetic development of CSS is presently beyond the scope of this paper, we nevertheless outline in short its development, based on Heinz Werner's (1978, p. 108–109) orthogenetic principle of development, that “wherever development occurs it proceeds from a state of relative globality and lack of differentiation to a state of increasing differentiation, articulation, and hierarchic integration.” This orthogenetic principle has been shown to be consistent with the genetic organization of the cortex (Chen et al., 2012). Akin to Werner's (1978) notion of increasing differentiation and hierarchic integration, CSS is proposed to manifest with development as a successively more complex structure. In support, Anokhin et al. (2000) report that EEG dimensional complexity increases with age between 7 and 17. Moreover, the two trajectories in the CSS should, early on in development, be indistinguishable (Werner's “relative globality”), and the corresponding CSS space should comprise one global sphere (and not two). Support for this proposition was given by a recent fMRI study showing that it is only from around 2 years that the antagonistic behavior between the cortical networks is first observed (Gao et al., 2013).

There could also be states where both trajectories are under the threshold for access awareness, for example dreaming, and states where both trajectories enable access awareness, where one attends to the activity of the intrinsic system, without being immersed in it, as an observer such as in meditation. Neuroscientific studies of the neural space support this intuition: during dreaming, most of the DMN deactivates, as well as the extrinsic system (Nir and Tononi, 2010). Similarly, activity in the intrinsic system may persist in parallel to extrinsic stimulation if external stimulation is not sufficiently challenging (Greicius and Menon, 2004; Wilson et al., 2008), or when one attends to the activity of the intrinsic system (Christoff et al., 2009), as is the case during meditation (Travis and Shear, 2010). Next, we describe cases of alteration in typical CSS dynamics. All these states, we suggest, involve an alteration in the regular sense of NS, as is the case in early childhood (Oatley, 2007).

We suggest that alterations in typical CSS dynamics occur when the regular sense of NS is modified. While the typically antagonistic behavior of the trajectories essentially indicates differentiation, here we discuss those conditions wherein hierarchic integration is achieved (in Werner's terms). These states have been termed “no-self,” “transpersonal,” or “transcendent” states, and they can occur in a spontaneous or training-induced manner (Alexander and Langer, 1990; Pascual-Leone, 2000; Hartman and Zimberoff, 2008). Such states have been related to enhanced performance (Csikszentmihalyi, 1988, 1990; Leary et al., 2006) and heightened happiness (Dambrun and Ricard, 2011). We predict that these situations can be seen phenomenologically (first two predictions) as well as being translated into neural space (last two predictions), resulting in:

To support these propositions, we bring evidence from two distinct states, both described as involving alterations in the sense of self: optimal experience, also called flow (Csikszentmihalyi, 1988, 1990), and meditation (Hölzel et al., 2011; Fell, 2012; Vago and Silbersweig, 2012). These two states have been sometimes considered to be largely similar (e.g., Kristeller and Rikhye, 2008; Bermant et al., 2011). However, others consider flow and meditation to diverge in their phenomenology and ultimate aim, as flow fosters development through higher challenges and skill refinement while meditation mainly points toward self-transcendence (Delle Fave et al., 2011). Moreover, flow is largely spontaneous and transitory, whereas the meditative state is training-induced and can become an enduring condition. While we do not expect isomorphism between flow and meditation in the phenomenological and neural spaces, we bring both as examples of an alteration in the sense of self and increased hierarchic integration, demonstrating how we can put the CSS to work.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.