Contribution of animal research to the search for the neural correlates of
consciousness
Number of EC: 10
Name: Laury van Bedaf
Student number: 5931630 Daily supervisors: dhr. G. Klein
Co-assessor: Prof. Dr. C. Pennartz Research institutes: SILS-CNS
Project: Literature thesis
Education: MSc in Brain and Cognitive Sciences, track cognitive neuroscience at the University of Amsterdam
Word count: 9966
Abstract
Consciousness has been a subject of interest for many fields for many years but as yet there is no consensus upon the definition of consciousness. Three main theories of consciousness will be discussed: integrated information theory, recurrent processing and the global
neuronal workspace model. V1 and the fronto-parietal network are found important for human consciousness therefore there is focused upon these areas. These areas are comparable between humans, monkeys and rodents. The paradigms; blindsight, flash suppression, figure ground modulation and the sustained attention task are often used for research related to consciousness in rats and monkeys. This paper supports the idea that to search for the neural correlates of consciousness and further understand human
consciousness animal models should be studied as animal models allow for invasive in vivo research.
Definition of consciousness
Consciousness is a concept which has been studied by many different fields for many years. One of the first to study consciousness was the Greek philosopher Aristotle who thought about the perception of what we perceive in his book ‘On the soul’ (c.350 BC). The central question was are you conscious of your consciousness? Another early idea about
consciousness is dualism from Descartes. In his view, mind and matter are distinct and cannot be reduced to a single concept. “Matter” in this context is analogous for the body. The mind is a nonphysical substance and related to consciousness but it is distinct from the brain. The brain was believed to be the seat of intelligence (Descartes, 1641 (trans: 1984); Robinson, 2003). Thus, in this view consciousness is distinct from matter.
Despite the many years of research neither philosophy nor science has been able to give a good definition of consciousness. Some scientists (Crick & Koch 1998) do not give a clear definition of consciousness, because the lack of knowledge about consciousness increases the risk of making an incorrect definition. This could lead to more confusion than clarity. Many phenomena can fall under the definition of consciousness such as: the stages of consciousness (i.e. vigilance, wakefulness, sleep, coma or anesthesia), higher order forms of consciousness such as meta-consciousness, self-consciousness, higher order thought and thinking itself or free will.
Consciousness in this thesis is in line with Tononi’s suggestion (2004, 5:42): ”Think of it as
what abandons us every night when we fall into dreamless sleep and returns the next
morning when we wake up. Without consciousness there would be neither an external world nor our own selves: there would be nothing at all”. A narrower form of this consciousness is
discussed in this thesis; the conscious experience of a (visual) event.
Within neuroscience, there is no consensus upon the definition nor the concepts of consciousness of experience. Concepts which surround consciousness are: attention, reportability, integration of different features, emotions, memory and qualia of an experience. The importance of these concepts differs from theory to theory.
For example Lamme (2006) promotes the idea that to find the true neural correlates of consciousness we have to stop searching for these correlates as an exact representation of our instinctive view of consciousness. We have to let neuroscience lead the way to find a
new and good definition of consciousness. To do so we have to start to make a clear distinction between different processes. This means that reportability, language, attention, speech and memory are not consciousness.
Dehaene (2006) made a taxonomy of consciousness to better grasp the concept with two factors: attention and stimulus strength. In his view being conscious of something only happens when attention (bottom-up or top-down) is given to the stimulus and the strength of the stimulus itself is high enough. This definition of consciousness also gives 3 forms of unconscious processing. First subliminal processing: subliminally presented stimuli are not perceived consciously because the strength of the stimulus is not high enough although attention is paid. Second is pre-conscious processing: the stimulus strength is high enough but no attention is paid, for example because the subject is distracted. The third form is lacking in both; the stimulus strength is not high enough and no attention is paid. In summary, Lamme proposes that consciousness can emerge without attention and
Dehaene proposes that attention is needed for information to become consciousness. Due to these kinds of differences it is difficult to identify the neural correlates of consciousness. In the next section I will give an overview of the main ideas of consciousness as understood within neuroscience; the integrated information theory, recurrent processing and the global neuronal workspace model.
Integrated information is consciousness
Tononi (2004) proposes that consciousness is information integration and has a dynamic core. Tononi divides consciousness into two different problems. The first problem is the quantity of consciousness which is defined by the number of functional connections between elements. These connections make different complexes. Within these complexes the information is integrated. This is crucial for consciousness because a conscious
experience is always integrated (Tononi & Edelman 1998) One can only have one conscious experience at a time. This is, for example, shown with binocular rivalry, where only one image can be perceived at a single moment in time. Tononi(2004) proposes that the
complexity of the complex wherein the information is integrated can be measured with phi. The leads to a graduate scale of consciousness, for example the system with the highest phi has the “most” consciousness. In sum, the amount of integration of information makes the quantity of consciousness.
The second problem is the quality of consciousness, which is related to the hard problem of consciousness (Chalmers 1995a, 1995b). If the integration of information gives rise to consciousness, what gives us the different conscious experiences? For example, what is the difference between seeing red and blue? Tononi (2004) proposes that the organization of the integrated system defines the nature of the conscious experience. Each element which is currently in the system, and each element which is not, defines that specific conscious experience. So, if one is experiencing blue, all the neurons coded for blue (which are in the system) and also the neurons for seeing red (which are not in the system) are important for one’s experience of blue. Removing these neurons will change the conscious perception of blue. This is the quality of consciousness.
Quality and quantity together give rise to consciousness. Because consciousness is
dependent on the complexity of a system and the nature of this system, consciousness does not have a static but a dynamic core. This dynamic core gives us the possibility to perceive consciously an infinite number of different experiences. The neurons currently in the dynamic core have strong functional connections with each other. Some neurons can be in the dynamic core for one conscious experience but not for another (Tononi & Edelman, 1998). Thus, unique conscious experience is created by the current dynamic core of
consciousness. A key node for the dynamic core could be the thalamocortical system (Tononi & Edelman, 1998; Tononi 2004), which is known to be important for vigilance. The amount of integration and the complex in which this is integrated give rise to a specific conscious experience. This is the dynamic core of consciousness.
The integrated information theory is a mathematical and theoretical model of consciousness. To validate the theory, connections in the brain need to be tested to determine whether a group of neurons is excluded from the network due to hard-wired constraints or because of functional changes. This might be possible, in vivo, with electrophysiological recordings in animals. Furthermore, the model has good explanatory value for disorders of consciousness such as split-brain, paradigms often used in psychology such as attentional blink and
binocular rivalry (Tononi 2004) and for phenomenal dreams (Nir & Tononi, 2009).
Furthermore, there have been attempts to test the quantity of integrated information in an artificial agent but no clear results have been found (Edlund et al., 2011). If this is possible the artificial agent could serve as a model to test the integrated information theory. Recurrent processing
Recurrent processing is proposed by Victor Lamme to be the neural correlate of
consciousness. Visual information enters the brain through the retina then this information is moved “up” the hierarchy through the early visual areas to, for example, the prefrontal cortex. This is called the feedforward sweep. The feedforward sweep extracts meaningful features from the visual scene and prepares potential motor response to act on the incoming information. No consciousness has yet emerged (Lamme 2000). Then recurrent processing (RP) starts. This is feedback and interaction from the higher areas to the lower areas, enabling a widespread exchange of information (Lamme 2006). RP also elicits synaptic plasticity processes which are thought to be the neural correlates of learning and memory (Singer 1995). This RP might be the neural correlate of consciousness (Lamme 2006). Note; feedback is not a linear process and when the feedforward sweep arrives at V2 feedback has already begun. Recurrent processing is divided into local RP and global RP. The global RP, which involves the entire brain, is needed for reportable consciousness. This does not mean that a subject cannot be conscious of stimuli when local RP is happening only in the visual areas, the subject is just unable to report it.
The ideas of RP can be tested in many different set-ups with EEG and fMRI, for example with inattentional blindness. In inattentional blindness the subject’s attention is averted so that a normally visible stimulus is rendered invisible. During an inattentional blindness paradigm there is local RP in early visual cortex, but the RP does not extend to frontal brain areas (Scholte et al., 2006). The recurrent processing theory suggests the subject processed the stimulus, as there was local RP, but due to inattention he is not able to report it. Lamme (2006) proposes that after recurrent processing of information, that information is conscious although maybe not reportable.
The theory of recurrent processing can also explain the backward masking experiment. In this experiment a stimulus is hidden by presenting a mask quickly after stimulus. The
feedforward sweep is not disrupted and can reach motor areas (Deheane et al., 1998). When the information is fed back to the early visual areas, other information is already present in that area (e.g. the mask). This leads to a clash of the recurrent information for the stimulus and the present information from the mask then the subject does not become conscious of the first object.
Furthermore, RP can be tested with transcranial magnetic stimulation (TMS). These studies (Koivisto et al., 2011; Boyer et al., 2005) revealed that a magnetic pulse at ± 80 ms disrupt
the entire information flow, and therefore probably disrupt the feedforward sweep. A magnetic pulse at ± 120 ms led to decreased visual awareness but no decreased
performance in a forced-choice paradigm, thus this pulse might disrupt recurrent processing. These studies give evidence for two temporally different information flows (feed forward en feedback information flows). Other forms of evidence for RP can be found in figure ground modulation and blindsight, which is discussed later.
Global Neuronal Workspace Model (GNW)
This model implies that information which is “broadcasted” in a conceptual global workspace becomes conscious. A lot of information is available in the brain but only a selected amount of information is broadcasted (Dehaene & Changeux, 2011; Dehaene et al., 2006; Dehaene 2001). Visual information first enters the cortex through the early visual areas from here it is moved up the hierarchy through a bottom-up manner through neighboring areas. This information can access the global neuronal workspace if it is selected by attention. If it is not selected by attention the information stays pre-conscious. The information in the global workspace becomes self-sustained and global, mediated by gamma-band oscillations
(Dehaene 2003). Higher areas are synchronized when a stimulus is broadcasted in the global workspace. This global and exclusive availability for a given stimulus is what we experience as consciousness. Information in the global workspace is available for further processes such as memory, evaluation and verbal report. This implies that the competition of information to become conscious is a competition held before the global workspace. The information which receives the most attention will access the global workspace.
Although there is no exclusive evidence which proves the global neuronal workspace model, there is a lot of evidence which points in that direction (Dehaene & Changeux, 2011). For example, fMRI masking experiments give a contrast between the subliminal and consciously perceived stimulus because the stimulus strength differs. Differences between seen and unseen stimuli are found in the extrastriate cortex, parietal cortex and the prefrontal cortex (Dehaene et al., 2001; Haynes et al., 2005b).
The attentional blink paradigm, changes blindness, and inattentional blindness make it possible to dissociate between preconscious and conscious processing because in both conditions the stimulus has enough strength to become conscious but this does not happen as top-down attention is diverted. Differences between preconscious and conscious
processing are found with fMRI studies (Marois et al., 2004) and EEG studies (Sergent et al., 2005). These differences occur 200-300 ms after the stimulus in the frontal-parietal areas and are are characterized by amplification of posterior activity. For a clear overview of the experiments conducted which support Dehaene’s global workspace hypothesis refer to the review paper of 2011 (Dehaene & Changeux). The difference between conscious and unconscious processing, in Dehaene’s view, is visible throughout the entire cortex and especially in the higher brain areas.
Comparing the theories
All these theories describe the same phenomenon and all agree that when one can report a stimulus, this stimulus is perceived consciously. For this, information moves from the visual areas up the hierarchy to the fronto-parietal areas and when consciousness emerges there is widespread synchronization. During widespread synchronization (mainly in the beta and gamma band, Dehaene & Changeux, 2011) clusters of neurons are highly interactive, therefore information can be integrated. Then there is a late amplification of the relevant sensory activity and large-scale ignition of the fronto-parietal network (Dehaene & Changeux, 2011). Where the GNW and RP differ is when one consciously perceives.
Dehaeane et al. (2003b) state that one becomes conscious of the stimulus if it is occupying the global workspace and then the information is available for report. Whereas Lamme (2006) proposes that local RP is enough for consciousness although it is not necessary that one can report it. Furthermore, Dehaene (2003b) states that attention is needed to become conscious, whereas Lamme (2003) proposes that RP alone is sufficient for consciousness. Thus there is disagreement on reportability and attention.
The disagreement when one become conscious is not found when consciousness would be measured on a gradual scale, as with phi. Then local RP can be conscious, only to a lesser extent than when attention is paid and the stimulus is available for report.
The disagreement is also solved by making a distinction between phenomenal and access consciousness (Block, 2005). Phenomenal consciousness is best described in the difference between redness and greenness and has great similarities with (local) RP (Lamme 2006) and the quality of consciousness (Tononi 2004). Access consciousness is the information which is accessed or is ‘broadcast’ in the global workspace (Block, 2005; Block, 1990). Both these
processes are a part of consciousness because both give rise to the richness of the content of our consciousness, for example the experience itself when one sees a colour or scene.
Despite many years of effort the research is not moving forward as fast as hoped. To move forward more experimental evidence is needed. However, human experiments are limited. EEG studies have good temporal resolution but are limited in their spatial resolution. fMRI studies have reasonable spatial resolution but have bad temporal resolution. Furthermore, these are indirect measurements of brain activity. Only a few studies (e.g. Quiroga et al., 2008) have been able to measure directly from the human brain. These studies are always limited in the freedom of measuring areas and population. Therefore consciousness research might benefit from animal models where invasive research is possible. In animals, neurons can be activated or inactivated and direct neuronal activity can be measured with
electrophysiology from neuronal assemblies from different brain areas simultaneously (Carandini & Churchland, 2013). Consciousness research already uses monkeys but rats are also a possible candidate. In order to study consciousness in rodents it must be established how far the anatomy and function of the relevant brain areas in humans, monkeys and rats are similar.
Anatomy
Humans and monkeys are both primates and share a common ancestor about 25 million years ago. Furthermore humans, monkeys and rats are all mammals and share a common ancestor about 75 million years ago (Carandini & Churchland, 2013). Therefore we share fundamental similarities in the brain organization (Carandini & Churchland 2013, Krubizet 2007). Areas which are found important for consciousness in human studies are the visual areas, the posterior partial cortex (PPC) and the prefrontal cortex (PFC).
Visual system
Our understanding of the visual system is mostly based of electrophysiological studies on macaque monkeys and cats. In humans as well as in monkeys 90% of the visual information reaches the brain through one main route. The information is perceived in the retina, from the retina the information moves up to the lateral geniculate nucleus (LGN) to V1, and from V1 to the rest of the brain (Felleman & Essen 1991; Esssen et al., 1992). Beside bottom-up
input from the LGN, V1 also has afferent input from extrastriate visual areas including, V2, V3, V4, MT, PO, PIP, TEO and TE, and also from subcortical areas such as the inferior pulvinar, the amygdala and the claustrum (Leopold 2012). The connections in the visual cortex are mostly reciprocal but V1 also receives direct information from areas to which is does not directly have feedforward connections. These areas include MST, FST, STP, LIP, frontal eye fields and the auditory cortex (Tong 2003).
The visual cortex is built up in a hierarchical manner. Cells in the retina and LGN have very small receptive fields. From V1 up to higher areas the receptive fields increase and when moving up, more complex scenes can be processed. V1 neurons are sensitive to orientation (Knierim et al., 1992), motion direction (Snowden et al., 1991), colour contrast (Conway et al., 2002) and the neurons are still ocular dominated (Cumming et al., 2002; Hubel et al., 1968). Due to the very small receptive fields, it has a very high spatial frequency (Tong 2003). After V1 the information is passed to the dorsal and ventral stream pathways. The ventral pathway includes the inferior temporal cortex and it is thought to process form recognition and object representation. The dorsal stream includes the dorsomedial areas and the
posterior parietal cortex and is associated with motion, representation of object location and motor planning.
The remaining information is projected from the retina to various other subcortical structures, and in this way bypasses V1. One important parallel pathway is the retina – superior colliculus – pulvinar. The pulvinar has reciprocal connections with several extrastriate areas (Leopold 2012).
Rats’ vision has long been underestimated. Although the acuity of the rat (on average 1.0 cycles/degree) is still much less than for humans, rats are used in visual guided tasks and perform well (Prusky et al., 2002). So, it might also be possible to use rodents for visual awareness research. There is a big difference in vision between different breads of rat, thus the breads must be selected carefully. The Fisher-Norway rats have the best acuity, although the acuity is also dependent on the nature of the visual experience during development (Prusky et al., 2002).
The mouse’s visual system shares fundamental organizational principles with other species, for example the visual cortex is divided into functionally and structurally different areas. It is more highly developed than was expected and there is evidence for a dorsal and ventral
stream in mice (Marshel et al., 2011). Furthermore, neurons in rats’ V1 have diverse visual properties, necessary for perception (Girman, 1999) and the response properties are
comparable between primates and rodents (van den Berg, 2010). In general the rodent visual system shares many of the complexities of a primate visual system (Wang & Burkhalter, 2007).
Posterior parietal cortex
The posterior parietal cortex (PPC) is divided into several functional subdivisions: the medial temporal area, the medial superior temporal area, the medial lateral intraparietal area, and the lateral intraparietal area (LIP). Most monkey research focuses on the LIP (Cohen, 2009). LIP receives visual (from the extrastriate visual area, cortical areas, limbic areas and
brainstem areas) and auditory input (direct input from the tempo-parietal cortex) (Cohen, 2009). The PPC of rodents is very similar to the LIP (brodmann area 7) in monkeys. It has connections with the thalamus, the auditory cortex and the medial agranular cortex and the orbital areas (Reep et al., 1994). Thus both primates’ LIP and rats’ PPC receive input from visual and auditory areas.
PPC neurons in rats can reflect information pertaining to the current position in space
(McNaughton et al., 1994; Snyder, 1998), the direction of motion (Chen et al., 1994), the type of locomotor behavior (McNaughton et al., 1994) and integrate this information so that it can be applied to multiple spatial settings (Nitz, 2012). LIP neurons are found to be important for the planning of eye movement (Ipata, 2006) and hand movement (Kalaska, 1996).
LIP neurons in monkeys and PPC neurons in rats are found to react to stimuli in their
receptive field regardless of the features of the stimulus (Thomas & Pare, 2007; Broussard et al., 2012) but especially to relevant stimuli (Suzuki & Gottlieb 2013; Broussard et al., 2006). Furthermore PPC is needed to shift attention (Bisley et al., 2004; Ipata et al., 2006). After a lesion in PPC, humans (Heilman et al., 1983), monkeys (Rizzolatti et al., 1983) and rats (Reep et al., 2004) show similar symptoms; subjects lack spatial and personal awareness
contralateral to the lesion .This is called neglect. This similarity is evidence that the PPC serves the same function across these species (Reep et al., 1994; Broussard, 2012) and that human’s, monkey’s and rat’s PPC is comparable.
PFC is located rostral to motor and premotor areas and is divided into several sub-areas. In primates PFC consists of ACC, dlPFC, mlPFC (Seamans et al., 2008). Rats’ PFC is divided into the anterior cingulate, the prelimbic(PL) and the infralimbic cortex (Uylings et al., 2003). For both primates and rats the PFC has connections to sensory cortices, premotor cortices, somatosensory cortices and limbic areas including the amygdala, perirhinal and entorhinal cortex (Onger and Price, 2000). Furthermore it is part of the PFC-basal ganglia-
thalamocortical circuitry (Uylings et al., 2003; Heidbreder & Groenewegen, 2003) and of the cortical-prefrontal-basal forebrain circuitry (Golmayo et al., 2003). Rats’ rostral part of the cingulate cortex reacts to visual cortex stimulation and an area dorso-lateral adjacent to the precentral-motor association area responds to the stimulation of the somatosensory cortex (Golmayo et al., 2003). Thus there are cross species similarities based on connectivity (Uylings et al., 2003).
Similarity based on function is difficult to establish because PFC is thought to be important for higher mental processes. Lesions studies in humans (Shallice, 1982; Robbins et al., 1996) and monkeys (Fuahashi & Kubota, 1994) reveal that PFC lesions affect the manipulation of information in short-term memory to achieve a specific goal, the ability to use memory to plan events and to predict upcoming events. Furthermore several studies have indicated that top-down influences originate from the prefrontal cortex (Posner and Petersen 1990; Fuster 2001; Miller and Cohen, 2001; Gregoriou et al., 2009; Orr and Weissman 2009; Totah et al., 2012).
The sub-areas of the PFC probably have distinct functions although what these functions are is still under debate. The dlPFC is active in every task preformed by trained monkeys and rats (Seamans et al., 2008). Suggested specific functions for the dlPFC are abstract rule learning (Freedman et al., 2001), rule flexibility (Rich and Shapiro, 2007) and attention (Muir er al. 1996; Broersen& Uylings, 1999). The dmPFC is thought to be important in controlling the timing of an action (Narayanan & Laubach 2009). Frontal areas are also involved in
proactively inhibiting a response during the delay period until it is the right moment for the response (Boulinguez et al., 2008).
The function of the ACC can be very diverse. For example, lesion studies in rats (Seamans et al., 1995) and monkeys (Rushworth et al., 2003; 2007) revealed that lesioned animals do not act on the predicted reward anymore. Subjects with ACC-lesions have deficits in reactive adjustments during a task (Pellegrino et al., 2007). In general it can be said that the ACC is
important for voluntary response selection (Deiber et al., 1991; Petersen et al., 1988) and tracking and monitoring the events of a task (Veen & Carter 2002; Seamans et al., 2008). Thus, although the function of the PFC stays unclear, lesion studies indicate that there are function similarities across species. Based on connectivity and function, human, monkey and rat PFC is comparable.
In conclusion, for the discussed cortices there are great similarities between rats, monkeys and humans. Although translation across species is always complicated and needs to be performed with caution, the similarities show that these animals can be used as models in the search for the neural correlates of consciousness.
Do animals have consciousness?
Having established that the brain areas found important for consciousness in human research are comparable to those of monkeys and rats, this raises the question whether animals can also have consciousness.
In order to search for animal consciousness, it first has to be stated that the amount of consciousness could vary with the complexity of a system (Tononi 2004). This means that for a less complex systems (a rat compared to a human) consciousness can emerge but on a lesser level. It is difficult to give absolute evidence for animal consciousness, merely because an animal cannot report that it is conscious. Even if it could give a verbal report, as a human can, this is not exclusive evidence that the subject was not conscious of the not reported stimulus (Lamme, 2006; Lamme, 2010). An higher order form of conscious is required to access experiences in order to report what has been perceived. This could be
self-consciousness. Thus the absence of accurate reporting of consciousness does not imply the absence of consciousness (Edelman & Seth, 2009).
It is unlikely that all animals are without consciousness as consciousness is thought to be the most efficient way of reacting to a new and changing environment (Griffin & Speck, 2003). Moreover, Lovibond and Shanks’ review (2002) concluded that classical conditioning in humans (at least eyeblink and skin conductance conditioning) only happens when there is conscious recognition that the unconditioned stimulus is predicted by the conditioned stimulus. Lovibond & Shanks (2002) concluded that “there is little convincing evidence for Pavlovian condition without awareness” (p42). Conditioning is a well accepted and often
used technique in animal research (Griffin & Speck, 2003). Stating that animals do not have consciousness would lead to the conclusion that animals can complete a task without consciousness where humans need consciousness. Although the necessity of awareness in classical conditioning is still controversial, this is hinting at consciousness in animals. Other studies which might have evidence for animal consciousness are studies of episodic memory. For episodic memory one has to be able to travel back in time mentally and recollect specific past events which is called ‘autonoetic consciousness’ (Tulving &
Markowitsch, 1998). For a memory to be categorized as an episodic memory it has to have a what, where and when component and these components need to be integrated to form a flexible memory (Clayton et al., 2003).
The study of Hampton et al. (2001) shows that monkeys know what they remember. In this study one item was presented and the monkey had to be press a button to start the test. After a certain delay two items appeared and the monkey needed to touch the matching item. A correct response resulted in receiving the preferred food reward. After the monkeys had mastered this task a second option was given. They could choose not to do the test and receive an “unpreferred” food reward. With a longer delay between the first item and the start of the test, the monkey pressed the “do the test” button less often. If there was no sample array (catch trial) one monkey did not press the “do the test” and the other monkey pressed the button less often. Hampton et al. (2001) concluded therefore that the monkeys knew what they remembered. This might be evidence that monkeys have episodic memory but there is no ‘where’ component in this task. Furthermore one cannot exclude that the animal has learnt that it is better to opt out if the stimulus delay is long. To exclude this strategy different ways of degrading the memory trace should be used, such as variation in the stimulus duration and intensity (Shea & Heyes, 2010).
The episodic memory of Scrub jays was also tested (Clayton et al., 2003). Scrub jays could cache worms, the most preferred but perishable food, or nuts, less preferred but non-perishable food. They were allowed to cache food twice, in the morning or the next day. Depending on how much time there passed before reclaiming the food, scrub jays choose worms or nuts. For example, if the worms were cached first and therefore probably gone bad the scrub jays reclaimed the nuts. This indicates that scrub jays have episodic memory because they remember what is cached, where and when. Thus this is evidence that scrub jays can travel back in time mentally and can have episodic memory. This indicates that they
might have autonoetic consciousness as defined by Tulving. Although these two studies do not prove that these animals have consciousness it is pointing in that direction.
Although the studies describe above hint to the existence of consciousness in animals it is hard to prove because animals cannot verbally report being conscious. Concluding that consciousness can exist in animals is not necessary to let animal research contribute to the search for the neural correlates of consciousness. Consciousness can be studied by looking at elements which are needed for consciousness rather than looking directly at consciousness. These elements can also be found in animals as the anatomy of the areas is similar. So we will leave the question of consciousness in animals without answer and move on. With a neuroscientific approach, it is possible to look at objective neural activity for subjective consciousness. (Deheane et al., 2003, Lamme, 2005) and this leaves us with no reason to exclude animals from the search for the neural correlates of consciousness.
How can animal research contribute to the study of consciousness?
We have established that the cortices for rats, monkeys and humans are similar and that areas found important for consciousness in humans are also found in monkeys and rats. Thus how can consciousness be tested in these animals and how can this research contribute?
Blindsight
One of the first fields of research into human and animal consciousness is blindsight. “Blindsight is the ability of patients with clinically blind field defects, caused by damage to the primary visual cortex, to detect, localize and even discriminate visual stimuli that they deny seeing” (Cowey, 2010, p.3). In blindsight, vision is severely damaged and subjects report being blind, but visual guided behaviour is still possible. However the subject denies seeing objects on which the visual guided behaviour is based. This led to the conclusion that it is impossible to become aware of visual information after destruction of V1. This raises the question of whether V1 plays a key role in consciousness.
Lesions in human subjects are always arbitrary and it is hard to decide how much of V1 remains. Therefore the risk of testing humans with blindsight is that the found blindsight is based on artifacts (Cowey, 2010). For example, that the residual vision is based on an island
of surviving V1. To avoid this kind of artifact, good controlled experiments are needed. Monkey research can be of help. In monkey research V1 is damaged by a surgical lesion to ensure that no V1 activity is possible. Thus any residual vision which is found cannot be dependent on an island of surviving activity in V1. On order to test whether the lesions result in blindsight, a forced choice task and a detection task are performed. Humans with
blindsight can perform a forced choice task correctly but cannot carry out the detection task (Stoerig & Cowey, 1997).
In the forced choice task, the monkeys had to make an eye movement in response to a stimulus. The stimulus was shown in the blind field or in the good field. After training the monkeys were able to reach a 90% correct score (Cowey, 2004). This shows that the monkeys have visually guided behaviour.
In the second experiment the monkeys needed to categorize trials as blanks or as signal trials. The monkeys were trained to classify signal and blank trials presented in their good visual field. Later, in 10% of the trials the signal was shown in the blind field. If the signal was shown in the blind field the monkey categorized it as a blank whereas unlesioned monkeys classified these trials correctly. This indicates that the monkeys are not aware of the stimulus in their blind field (Cowey, 2004) although they perform well on a forced choice task. Thus monkeys can have blindsight.
Two additional experiments showed that these results were not confounded by light
scattering into the good visual field (Cowey, 2010) nor that the monkey became aware of the stimulus with other senses (Cowey, 2004). Thus we can conclude that the loss of V1 leads to the loss of awareness while some visual guided behaviour remains.
V1 is the main gate for all visual information. To have any kind of visual behaviour after V1 damage, the information needs to by-pass V1. A likely alternative route to by-pass the LGN, V1, V2 is that the information comes directly from the superior colliculus and pulvinar to higher areas such as MT (Lamme & Roelfsema, 2000). Which precise alternative route is used in blindsight is unknown (Leopold, 2012) but it is sure that visual information reaches higher areas because there is still activity found in extrastriate cortices, especially in MT(Bridge et al., 2010; Bruce et al., 1986).
How is it possible that no awareness can occur? This again indicates a special role for V1 in consciousness. The most conservative idea of the role of V1 in consciousness is that V1 is a gate for almost all the visual information passing to the brain. So, after this gate is destroyed,
almost no information can reach the brain. Therefore the activity in those areas is not sufficient to evoke awareness (Leopold, 2012). An alternative view is that because V1 is destroyed, recurrent processing is impossible and therefore the information cannot become conscious. This indicates that V1 is more than the main gate for the visual information to enter the system. Evidence for this is that visual cortical neurons remain active after their participation in the feedforward sweep (Lamme & Roelfsema, 2000; Lamme, 2001) and the neurons change their tuning characteristics over the course of their response, as is seen in the figure ground modulation(discussed later). Recurrent processing is seen ~100 ms after the start of the feedforward sweep, and this is what gives rise to consciousness .Thus in blindsight, information can reach higher through the alternative pathway but feedback to V1 is impossible (because there is no V1). Feedback can still occur to other areas surrounding V1 but apparently this is not sufficient for awareness. This further implies that V1 has a specific role in awareness (Lamme, 2001). Yet is activity in V1 always correlated to awareness?
Flash suppression tasks
Besides blindsight research, consciousness is also studied in monkeys using visual perception tasks in which the visibility of the stimulus is based on the monkey’s subjective report. A frequently used visual perception task is the flash suppression paradigm. In the flash suppression paradigm, a salient stimulus is shown on the screen. After a short delay a mask of randomly moving dots appears which can make the stimulus invisible. By changing the properties of the mask, the stimulus can be made visible or invisible. This visibility is based on the subjective report of the subject, human or monkey (Leopold et al., 2003).
During single unit-recording in V1, no difference in firing rate was found between a visible and a suppressed stimulus. However a difference was found between a suppressed stimulus and an absent stimulus. This indicates that V1 neurons represent the stimulus regardless of the subjective report of the monkey (Gail et al, 2004, Keliris et al., 2010). The opposite is found with fMRI (Polonsky et al., 2000), where the human primary visual cortex follows the subjective report. In an attempt to resolve this contradiction Maier et al. (2008) measured the local field potential (LPF) and the blood oxygen level dependent (BOLD) response from the same patch of tissue in two monkeys during the flash suppression task. The results were in line with the results found previously. The spiking and high-frequency power in V1 was not affected by the subjective report of the monkey. The BOLD response does resemble the
subjective report (Maier et al., 2008). This contrast is also found during binocular rivalry between fMRI (Tong, 2001), EEG (Lansing, 1964) and electrophysiological data in monkeys (Leopold & Logothetis, 1996). These studies show that a comparison between firing rate and BOLD response is difficult and contextually dependent. It might also reveal something about the different functions of these responses. Under normal conditions BOLD response and firing-rate resemble each other well, thus the existence of this contrast during suppression could possibly mean that BOLD reacts more on the feedback activity (Leopold, 2012) whereas single unit activity might be more driven by feedforward sweep. Why feedback activity would have a lesser effect on the signal unit activity is still unknown. In sum, signal unit activity recorded during flash suppression and binocular rivalry suggests that V1 activity is not correlated with the subjective report.
Figure ground modulation
Another visual perception paradigm often used with monkeys is the figure ground segregation task. In this task the monkey is presented with line segments over the entire screen. When the trial starts, a square comprising line segments in another direction may or may not appear. When a figure is presented, the monkey has to respond with an eye
movement otherwise the monkey has to remain unresponsive (Super et al., 2001).
Simultaneously, activity from V1 is measured. If a figure is presented in the receptive field of V1 neurons, the activity is enhanced compared to when the background was in the receptive field. This difference in activation is called the figure ground modulation (FGM). Moreover, the activity differs when the stimulus was seen as opposed to missed. When the stimulus was not seen the activity resembles the activity when there is no stimulus on the screen. FGM is therefore not a pure sensory process. By decreasing the salience of the figure, FGM was also found on trials where the stimulus was not seen. So FGM is also not the outcome of the decision process and seems to be an intermediate internal representation of the
stimulus. To further disentangle the function of FGM, the decision criterion was manipulated by changing the percentage of catch trials. By increasing the number of catch trials, and also the decision criterion, FGM was also present on not seen trials in the salient condition. While removing the catch trials in the less salient condition also removed the FGM in the not seen trials. It seems that only when the FGM exceeds the decision criterion a saccade is made and the decision criterion is dependent on the context like the number of catch trials. The
behaviour seems to be a combination of the FGM and a perceptual decision making process (Super, 2001).
FGM is seen in V1 but the receptive fields in V1 are small and the figure cannot be captured in one of these receptive fields, thus V1 neurons change their response properties based on contextual information occurring outside the classical receptive field (Knierim & van Essen, 1992, Kapadia et al., 1995). Moreover, most FGM is found in the cells which are at the edge of the figure (Poort et al., 2012). Thus FGM must be driven by higher areas which feed back to V1. This is further indicated by the late onset of the FMG, ~100 ms after onset (Lamme & Spekreijse, 2000). Therefore, FGM is a result of recurrent processing.
Molecular level of recurrent processing
To investigate the molecular basis for figure ground modulation and with that the molecular basis of recurrent processing in V1, an electrophysiological experiment with macaques is performed (Self et al., 2012). In this experiment monkeys perform a figure ground
modulation task. During this task, activity in V1 is measured and three different drugs which react on the glutamate receptors are administered; AVP, Ifenprodil and CNQX. AVP and Ifenprodil are blockers for a sub-type of the glutamate receptors; NMDA receptors. CNQX is a blocker for another subtype; AMPA receptors. This study reveals that glutamate receptors are important for information processing and that there is a difference between feedforward and feedback projections. The drug APV has no effect on the initial peak of the V1 activity but does reduce the FGM whereas ifenprodil increases the initial peak and also reduces the FGM. CNQX reduces the overall activity of the visual information (Self et al., 2012). Thus both NMDA receptors blockers have a negative effect on the FGM and the AMPA receptors
blockers have a negative effect on the total visual response.
It has been suggested that feedforward connections drive the neurons in the cortical column whereas recurrent connections can only modulate the activity (Crick & Koch, 1998). AMPA receptors are important for the overall visual information and the NMDA receptors are important for the FGM. This suggests that AMPA receptors are to procreate visual activity from lower to higher areas whereas NMDA receptors cause modulatory effects which mediate recurrent connections. This is in line with the notion that NMDA receptors can only depolarize the neuron when the magnesium block is removed from the receptor and the magnesium block is only removed when the cell is already firing. AMPA receptors mediate
neuronal firing for visual information. This suggests that NMDA effects can only be found after the activation of AMPA. NMDA receptors can increase the firing rate of active cells and have little effect on neurons which are not driven by the visual stimulus, in other words, recurrent connection can only affect neurons which have already been affected by the visual stimulus.
To summarize, NMDA receptors are important for FGM (Self et al., 2012) and FGM is
correlated with the detection of the stimulus (Super et al., 2003). The combination of these results suggests that recurrent interactions between V1 and high visual areas are necessary to solve the figure ground segregation task. The process would then be the following: first the AMPA receptors become active in the feedforward sweep; then the recurrent
connection, mediated by the NMDA receptors, can influence the firing rate but only for the neurons which are already activated by the feedforward sweep (Self, 2012). This recurrent activation causes the FGM and this results in the possibility of detecting the stimulus if the FGM exceeds the decision criteria.
In conclusion, activity in V1 is found to correlate with the subjective report based on evidence of blindsight, FGM studies, flash suppression and binocular rivalry paradigms. Yet the activity in V1 is not causal of the subjective report. Therefore V1 activity is probably important for awareness but not sufficient (Tong 2003). Rather, recurrent processing from higher areas to V1 is necessary for reportable awareness. These higher areas might be the fronto-parietal areas. The areas fusiform areas, inferior prefrontal, medial frontal and parietal cortex are found to differ in human fMRI masking studies. These studies make a contrast between consciously and unconsciously perceived words (Deheane, 2001; Haynes et al., 2005). Neural activity in these areas amplifies during conscious access (Deheane & Changeux, 2011). The following paragraphs will discuss what is found on the fronto-partial network in animal research related to consciousness.
Fronto-parietal network
First I will elaborate on how information is represented in the PPC and PFC. It has been hypothesized that the PPC can form a pure salience map. To test this, macaques were taught to perform a passive fixation task (Arcizet et al., 2011). Here there was one salient stimulus and 6 non salient stimuli. None of the stimuli had a behavioural significance for the task and
so the top-down influences were kept to a minimum. It was found that neuronal activity was higher for the salient stimulus. This indicates that LIP can form a pure salience map. The PPC forms this map by collating bottom-up salience signals from early visual areas. The map represents the stimulus characteristics and this salience map can be used to make a priority map. A priority map can be used to guide attention and saccades (Arcizet et al., 2011). Are we now able to conclude that activity in PPC for stimulus detection tasks is simply to make a salience map? The series of studies by Broussard et al., (2006,2009, 2010 and 2012) ascribe more functions to the PPC. All studies were performed with rats using the sustain attention task (SAT) or distractor SAT (dSAT). In the SAT the rats had to detect a brief and unpredicted signal light and had to reject trials where no signal was present. In dSAT a distractor light, which had to be ignored, was added (Bushnell et al., 1994).The rats had to press a lever to make a response (Bushnell et al., 1994). Increase in alpha band power predicted if the signal was correctly detected (Broussard & Givens 2010). Subpopulation of neurons in the PPC differed in activity between hits and misses, and increased when a signal was detected (Broussard et al., 2006). This response profile of the PPC for detected stimulus was independent of signal duration and moreover it was also independent of distractors. The distractors themselves were not represented in the PPC-neurons associated with detection. This indicates that the PPC does not represent the parameters ifself but that PPC neurons are selectively active during accurate signal detection of behaviourally-relevant signals.
Furthermore it is hypothesized that the PPC is involved in maintaining these signals for attentional processing to make a correct response among alternatives (Broussard et al., 2006). These results are in line with the monkey research by Thomas & Pare (2007) who also have shown that activity in the LIP does not represent the properties of the stimulus itself and that the activity is enhanced for relevant stimuli. Thus the PPC might be important for the detection of relevant stimuli.
It has been found that; the PPC can form a pure salience map of all stimuli, and that; the PPC selectively represents behaviourally relevant signals. In order to explain this contradiction between these results one has to look at the differences between the tasks. In order to find the pure salience map Arcizet et al. (2011) gave the monkeys a passive fixation task; there was no behaviourally relevant stimulus therefore the PPC could not represent this. Second, in the study of Broussard et al. (2006) and Broussard & Givens (2010), the distractors were given as a frequently flashing light. The light could have blended in with the background
noise and therefore it was not processed as a signal (Sarter et al., 2001) or detected by the detection-associated neurons. This might be the reason why this signal was not on the salience map. This is also in line with behaviour performance on the dSAT. The false alarm rate was raised but also the missed trials increased (Broussard et al., 2006) indicating that the task became more confusing and making it plausible that the distractor light blended in with the background noise. Whereas the distractors in the passive fixation task (Arcizet et al., 2011) were presented simultaneously with the salient stimulus.
Not only the PPC but the fronto-parietal network is important for selecting relevant stimuli. This is investigated in the study by Suzuki & Gottlieb (2013). Here monkeys had to make a saccade to the target when the “go” signal was given. During the delay a distractor could appear, which had to be ignored. The delay and spatial location of the distractor relative to the target varied over trials. During this task dlPFC and LIP were recorded. LIP neurons had a robust response to both the target and the irrelevant distractors whereas distractors in the dlPFC where more suppressed and distractor activity was negatively correlated with correct trials (Suzuki & Gottlieb 2013). Moreover inactivation of the distractor location in the dlPFC had a positive effect on the performance. This suggests that there is a strong inhibition of the visual distractors in the dlPFC. Furthermore reversible inactivation of the dlPFC and PPC revealed that inactivation of the dlPFC led to greater distractibility than inactivation of the LIP. Distractors which distracted the most were close the target location in space and time. This suggests that there were overlapping neuronal populations in dlPFC representing the target and the distractors and therefore it was more difficult to only suppress the distractor and this lead to more errors. Thus the dlPFC is important in suppressing irrelevant
distractors.
Difference in function of the dlPFC and LIP is further tested in the study by Buschman & Miller (2007). There, a pop-out task reflects strong bottom-up influences and a search task reflects strong top-down influences. The lateral PFC (lPFC), frontal eye fields (FEF) and LIP were measured while the monkeys preformed both tasks. This revealed that there is a difference in the order of activation between the three areas. During the pop-out task LIP became active first, about 30 ms for a saccade was made. Then the lPFC and FEF became significantly active. During the search task this was exactly the opposite. First the FEF and the lPFC and then the LIP became active, all before the eye saccade was made (Buschman & Miller 2007). Thus, both PPC and PFC are important for selection and perceptual processing.
dlPFC is thought to be the source of top-down attention. Whereas LIP/PPC can form a salient map of behaviourally relevant stimuli by integrating bottom-up information. This salience map is influenced by top-down attention to enhance activity of the target. The LIP might be capable of integrating top-down influences and bottom-up information and in collaboration with the PFC select the behaviourally relevant stimuli.
In order to explain how the PPC and PFC cooperate in the detection of stimuli and
suppression of distractors Broussard (2012) proposed that distractors increase activity in PFC which can recruit parietal acetylcholine (ACh) efflux in order to separate relevant and
irrelevant stimuli. This would lead to the network depicted in figure 1. During a well-learned and non-demanding task PPC’s main input comes from the PFC. If the rule changes or
distractors are added then the PFC recruits the basal forebrain cholinergic system (BFCS) and the locus coerules (LC). Through these areas the PFC mediates influx of ACh and
norepinephrine (NE), which are needed to suppress distractors (Broussard et al., 2012). Without ACh influx distractors are still being represented in the system. This result is also found for non-visual stimuli (such as auditory stimuli) (Berntson et al., 2003b). Thus with this network the PFC and PPC can effectively select the correct stimulus.
Figure 1: “Schematic diagram illustrating the main components of a neuronal network mediating the posterior parietal cortical (PPC) processing of relevant sensory signals.” (Broussard 2012 p. 4.).On the left side signals which are well-learned and on the right side signals when more distractors are added. Figure adapted from Broussard 2012 p. 4.
Research has shown that the fronto-parietal network is important in human consciousness. In animal research there is found that the PPC forms salience maps were top-down influence by the PFC can enhance the target location. Furthermore the PPC receives not only visual bottom-up input but also auditory input (Cohen, 2009) and can integrate top-down and
bottom-up influences from several modalities. The information from the PPC is used in the PFC to suppress distractors. Furthermore the dlPFC is also important in linking the stimulus to correct action and the timing of the response. This, in combination with the diversity of functions found for the PFC (see anatomy PFC), might indicate that this network acts like a global workspace (Seamans et al., 2008). In order to act like a global workspace the network needs to be very adaptable.
The neurons of ACC are adaptable. They can be involved in different aspects of the same task (Lapish et al., 2008). For example a task is divided into three stages; correct choice, reward and test phase. Neuron 2 (schematic depicted in figure 2) is involved in the correct choice phase and also in the reward phase but not in the last, the test phase. In the ACC specific neuronal assemblies are formed by strong recurrent excitatory connection to perform unitedly at specific stages of a task. Due to these assemblies the ACC can be highly adaptive and track the animal’s behaviour during different tasks (Lapish et al., 2008)
Figure 2: “Shown are three cell assemblies corresponding to correct choices (green), reward epochs (red), and basic test epochs (black). Neurons enclosed by the same colored outline are embedded within the same assembly, with neurons shared among assemblies being located within the intersections of the colored outlines. The firing rate of a neuron is indicated by its gray level. As the task progresses, first choice assemblies are activated (Lef), followed by reward assemblies (Center), followed by a general test phase assembly (Right). Depending on a neuron’s participation in various assemblies, numerous patterns of differential activity may be observed for single units, whereas at the same time, the cell assembly gives rise to a unique pattern at the network level” Lapisch et al. 2013 p. 11966. Figure adapated from Lapisch et al. 2013 p. 11966.
By forming neuronal assemblies, connections can be made on a neuronal level. Thus, the same areas and even the same neurons can be involved in different stages of the same task. This indicates a highly flexible and interconnected network which is needed for
Conclusion
Consciousness is a difficult and complex subject and different theories concerning its nature and its functioning coexist. This thesis is limited to the conscious experience of a visual event and is focused on how animal research can contribute to the search for the neuronal
correlates of consciousness. Human research is mainly focused upon the visual cortex and the fronto-parietal network, therefore these areas in animal research were the focus of this thesis. These areas are anatomically and functionally comparable between species. Thus animals could be used as a model to study consciousness and results found in animal research can contribute to understanding human consciousness.
The role of V1 in consciousness is studied in animals with different paradigms. Lesion of V1 results in blindsight, the inability to become aware of a stimulus. This indicates an important role for V1 in consciousness. In flash suppression and binocular rivalry paradigms the high frequency spiking rate is not correlated with subjective report whereas the BOLD response is correlated with the subjective report. The importance of V1 in consciousness might be found in recurrent processing. FGM is formed by recurrent processing and with a high salient stimulus FGM is correlated to the subjective report of the monkey. But FGM is an internal representation of the stimulus and therefore the FMG is not always correlated with the subjective report when the stimulus is not salient. Thus V1 is important for visual consciousness, probably because recurrent processing to V1 is necessary to become conscious of a visual stimulus. But V1 alone is not sufficient for reportable consciousness.
Information first accesses the brain through V1 then it moves up to the parietal areas. Animal research indicates that the parietal areas have the ability to combine bottom-up information and make a salience map of behaviourally relevant stimuli. This might indicate that this is the first area which makes a selection between all the presented information. Furthermore, the PPC is necessary to shift attention and thus can move the focus from one stimulus to
another. This selection is further enhanced by the dlPFC which is able to suppress distractors. Thus the bottom-up information and the top-down direction together select that stimulus upon which action will be taken.
It could be hypothesized that the parietal region might be the first area which selects the relevant stimuli. If a stimulus is selected by the fronto-parietal network this stimulus might gain access to the global neuronal workspace and then it is possible to remember, report and act upon this stimulus. For a system to react to an infinite amount of different stimuli with an infinite amount of different actions it needs to be an adaptable system. A system like this is found in the ACC in which different neurons can form neuronal networks to perform at specific stages of a task. This selection of the relevant stimulus might be attention and the availability for memory and action might be working memory. Thus the fronto-parietal network might be an important network of consciousness as describe in the global neuronal workspace model (Dehaene & Changeux, 2011).
This thesis has only focused on the visual cortex and the fronto-parietal areas. Other areas might also be important for consciousness. Which areas are important also depend on how consciousness is defined. When looking at consciousness from the GNW viewpoint, many different processes are involved, such as long-term memory, evaluating the value of the stimulus and the attentional systems to focus. All these system are also part of the neuronal correlates of consciousness. If recurrent processing is sufficient for consciousness, then the areas for that sense only could fulfil the requirements to become conscious. For example, for auditory consciousness, the auditory cortex might then be crucial.
To draw any conclusion about when consciousness emerges and what areas are involved more research is needed. Animal models could give valuable information about interactions and timing of different areas. With electrophysiological methods direct activity of groups of neurons in multiple areas (for example V1, PPC and PFC) can be measured simultaneously and this will give the information about timing of the different processes and interaction between these different areas. This method might also help to find neurological evidence for the integrated information theory. If different neurons can be measured while the animal is detecting a stimulus then the dynamic core of consciousness could be measured. Important insights are gained from animal research in, for example, the fronto-parietal network, but most animal research does not focus on consciousness. The research of Broussard et al. does to some extent researches consciousness because here the SAT is used. In this task the rat has to detect a signal. Failure in the detection task in subjects with blindsight is used as evidence that they are not conscious of the stimulus in the affected field and successful
completion of this task for stimuli in the non-affected field is evidence that the subject can become conscious of stimuli in that field. This evidence is generalized to blindsight research in monkeys. Why is it that if a rat can detect a signal this is not noted as the rat being conscious of that stimulus? Following this you might conclude that the studies by Broussard et al. are investigating consciousness and thus give evidence that the PPC is important for conscious detection of the stimulus.
In all theories of consciousness interactions and networks seems to be crucial. The study by Totah et al. (2012) investigated the communication between two separate areas within the prefrontal cortex: the prelimbic cortex (PL) and ACC. They recorded single unit spiking and local flied potentials simultaneously in rodents during SAT and found that there is synchrony between and ACC and PL within a wide range of frequencies. SAT was divided into two epochs. In the first epoch the animal had to focus on the areas where the signal was expected. ACC was then phase locked in beta oscillations to PL; it is suggested that this represents top-down attention. Beta synchrony was correlated with correct trials; a reduction in beta synchrony might suggest diminished communication between the areas and could therefore reflect an impaired cognitive state. The second epoch was the
preparation of instrumental action. Here ACC switched to delta oscillation and then retrained the PL. Delta oscillations are found in every trial indicating that it is important for making an instrumental action at the right time. The same neurons which participate in top-down attention now participate in the preparation for action. This difference in timing when the neurons are active gives crucial information and can only been found when separated neurons are measured. The temporal resolution of fMRI is far from capable for this. The temporal resolution of EEG might be good enough but the spatial resolution is too low to separate areas and far too low to separate neurons. Focussing this kind of research on consciousness in animals could be the key to move forward in the search for the neural correlates of consciousness.
In conclusion, consciousness is a very complex phenomenon and is probably generated by complex and interacting networks. Animal research can help to find and investigate these networks and therefore help to understand consciousness.
References
Arcizet, F., Mirpour, K., & Bisley, J. W. (2011). A pure salience response in posterior parietal cortex. Cerebral Cortex, 21, 2498-2506.
Aristotle, (c.350 BC). On the soul. Translated by J. A. Smith (1931), obtained from: http://ebooks.adelaide.edu.au/a/aristotle/a8so/complete.html
Berntson, G.G., Shafi, R., Knox, D.,& Sarter, M. (2003b). Blockade of epinephrine priming of the cerebral auditory evoked response by cortical cholinergic deafferentation.
Neuroscience, 116, 179–186.
Bisley, J. W., Krishna, B. S., & Goldberg, M. E. (2004). A rapid and precise on-response in posterior parietal cortex. The Journal of Neuroscience, 24(8), 1833-1838.
Block, N. (1990) Consciousness and accessibility. Behavior Brain Sciences. 13,596–598.
Block, N. (2005). Two neural correlates of consciousness. TRENDS in Cognitive Sciences, 9(2), 46-52.
Boyer, J. L., Harrison, S., & Ro, T. (2005). Unconscious processing of orientation and color without primary visual cortex. Proceedings of the National Academy of Sciences of
the United States of America, 102, 16875–16879.
Boulinguez, P., Jaffard, M., Granjon, L. & Benraiss, A. (2008). Warning signals induce
automatic EMG activations and proactive volitional inhibition: evidence from analysis of error distribution in simple RT. Journal of Neurophysiology, 99, 1572–1578.
Bridge, H., Hicks, S.L., Xie, J., Okell, T.W., Mannan, S., Alexander, I., Cowey, A. & Kennard, C. (2010). Visual activation of extra-striate cortex in the absence of V1 activation.
Neuropsychologia, 48, 4148–54.
Broersen, L.M., & Uylings, H.B. (1999) Visual attention task performance in Wistar and Lister hooded rats: response inhibition deficits after medial prefrontal cortex lesions.
Neuroscience, 94, 47-57.
Broussard, J. I. (2012). Posterior parietal cortex dynamically ranks topographic signals via cholinergic influenc. Frontiers in Integrative Neuroscience,6(32), 1-10.
Broussard, J. I., & Givens, B. (2010). Low frequency oscillations in rat posterior parietal cortex are differentially activated by cues and distractors. Neurobiology of Learning and
Memory, 94, 191-198.
Broussard, J. I., Karelina, K., Sarter, M., & Givens, B. (2009). Cholinergic optimization of cue-evoked parietal activity during challenged attentional performance. European Journal
of Neuroscience ,29(8), 1711-1722.
Broussard, J., Sarter, M., & Givens, B. (2006). Neuronal correlates of signal detection in the posterior parietal cortex of rats preforming a sustained attention task.
Buschman, T. J., & Miller, E. K. (2007). Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science, 315, 1860-1862.
Bushnell, P.J., Benignus, V. A., & Case, M.W. (2003). Signal detection behavior in humans and rats: a comparison with matched tasks. Behavioral Processes, 64, 121-129.
Bushnell, P.J., Kelly, K.L., & Crofton, K.M., (1994). Effects of toluene inhalation on detection of auditory signals in rats. Neurotoxicology and Teratology, 16, 149–160.
Carandini, M., & Churchland, A. K. (2013). Probing perceptual decisions in rodents.Nnature
Neuroscience,16(7), 824-831.
Chalmers, D.J. (1995a). Facing up to the problem of consciousness. Journal of Consciousness
Studies, 200-19.
Chalmers, D.J. (1995b). The puzzle of conscious experience. Scientific American, 90-100. Chen, L.L., Lin, L.H., Barnes, C.A., & McNaughton, B.L. (1994). Head-direction cells in the rat
posterior cortex. II. Contributions of Rat Parietal Neurons Track Route Progression 755 visual and ideothetic information to the directional firing. Experimental Brain
Research, 101, 24–34.
Clayton, N. S., Bussey, T. J., & Dickinson, A. (2003). Can animals recall the past and plan for the future? Nature reviews Neuroscience, 4, 685-691.
Cohen, Y. E. (2009). Multimodal activity in the parietal cortex. Hearing Research, 2009, 100-105.
Conway, B. R., Hubel, D. H., & Livingstone, M. S. (2002). Color contrast in macaque V1. Cerebral Cortex, 12, 915-925.
Cowey, A. (2004). Fact, artefact and myth about blindsight. The Quarterly Journal of
Experimental Psychology, 57A,577–609.
Cowey, A. (2010). The blindsight saga. Experimental Brain Research, 200, 3-24.
Crick, F., & Koch, C. (1998). Consciousness and neuroscience. Cerebral Cortex, 8, 97-107. Cumming, B. G. (2002). An unexpected specialization for horizontal disparity in primate
primary visual cortex. Nature, 418, 633–636.
Dehaene, S., & Changeux, J. (2011). Experimental and theoretical approaches to conscious processing. Neuron, 70, 200-227.
Dehaene, S., and Naccache, L. (2001). Towards a cognitive neuroscience of consciousness: Basic evidence and a workspace framework. Cognition 79,1–37.
Dehaene, S., Changeux, J., Naccache, L., Sackur, J., & Sergent, C. (2006). Conscious,
preconscious, and subliminal processing: a testable taxonomy. TRENDS in Cognitive
Sciences, 10(5).
Dehaene, S., Naccache, L., Cohen, L., Bihan, D.L., Mangin, J.F., Poline, J.B., and Rivière, D. (2001). Cerebral mechanisms of word masking and unconscious repetition priming.
Nature Neuroscience, 4, 752–758.
Dehaene, S., Naccache, L., Le Clec’H, G., Koechlin, E., Mueller, M., Dehaene- Lambertz, G., van de Moortele, P.F., & Le Bihan, D. (1998b). Imaging unconscious semantic priming, Nature, 395, 597–600.
Dehaene, S., Sergent, C., & Changeux, J. (2003). A neuronal network model linking subjective reports and objective physiological data during conscious perception. Proceedings of
the National Academy of Sciences, 100(14), 8520-8525.
Deiber, M.P., Passingham, R.E., Colebatch, J.G., Friston, K.J., Nixon, P.D., & Frackowiak, R.S., (1991) Cortical areas and the selection of movement: a study with positron emission tomography. Experimental Brain Research, 84, 393-402.
Descartes, R. (1984–5). The Philosophical Writings of Descartes, translation. J. Cottingham, R.Stoothof, and D. Murdoch (2 vols). Cambridge: Cambridge University Press
Edelman, D. B., & Seth, A. K. (2009). Animal consciousness: a synthetic approach. Trends in
Neurosciences, 32(9), 476-484.
Edlund, J. A., Chaumont, N., Hintze, A., Koch, C., Tononi, G., & Adami, C. (2011). Integrated information increases with fitness in the evolution of animats. PLoS Computational
Biology, 7(10), 1-13.
Felleman, D. J., & van Essen, D. C. (1991). Distributed hierarchical processing in the primate cerebral cortex. Cerebral Cortex, 1, 1-47.
Freedman, D.J., Riesenhuber, M., Poggio, T., & Miller, E.K., (2001). Categorical
representation of visual stimuli in the primate prefrontal cortex. Science 291, 312-316.
Funahashi, S., & Kubota, K. (1994). Working memory and prefrontal cortex. Neuroscience
Research, 21, 1-11.
Fuster, J.M. (2001) The prefrontal cortex--an update: time is of the essence. Neuron, 30, 319-333.
Gail, A., Brinksmeyer, H.J., Eckhorn, R. (2004). Perception-related modulations of local field potential power and coherence in primary visual cortex of awake monkey during binocular rivalry. Cerebral Cortex, 14, 300–13.
