CODAM model: Through attention to consciousness
CODAM is an acronym for the much lengthier title 'COrollary Discharge of Attention Movement'. This longer title contains the crucial element of the model: it considers the so-called 'corollary discharge' or copy of the control signal causing the movement of the focus of attention to be the crucial element both in the general deployment of attention and during the creation of consciousness. The employment of a copy of a control signal is well known in control theory to lead both to speed-up and to increased accuracy in a general control systems. It is proposed in CODAM as a new component in models of attention, making attention more effective. A bonus of this extra component is however, not inconsiderable: it can help explain the creation of the inner self as the owner of an about-to-be-experienced stimulus which is being attended to. This process of 'attention causing consciousness' takes some non-zero time (some hundreds or so of milliseconds in the brain) and so a copy of the attention movement control signal is very important in protecting and speeding up this process. Such a control system view of the creation of consciousness thus can help make progress on the function of consciousness, a very controversial topic.
In order to make the above statements understandable we need to consider the appropriate features of attention as a control system in the brain. We can thereby develop a refined enough model of attention control, using the notion of an attention copy signal, to help understand some of the more intriguing paradigms considered recently. With this support for the CODAM model, we can then begin to make inroads into understanding the 'inner' or 'pre-reflective' self of the phenomenologists (Zahavi, 1999) and of P-consciousness of (Block, 1995).
It is usual to distinguish between bottom-up or exogenous attention (arising from the attraction of a salient stimulus requiring attention) and top-down or endogenous attention (such as occurs when searching for a given face in a crowd). Recent brain imaging data indicates considerable overlap in the brain between these two forms of attention. There is also the highly relevant observation of early post-stimulus prefrontal activation to a salient stimulus (Foxe & Simpson, 2002)). We propose thus to combine the two forms into one, for which there is an overall goal to attend which has been created some time in the past (for the endogenous form) or just created 'on the fly' by a salient input (in the endogenous form). When we discuss attention in general either of these two forms are meant.
Attention is an important gateway to consciousness. Without attention it would seem that one cannot be aware of an object in the external world. The phenomena of the attentional blink (Vogel and Shapiro 1998), inattentional blindness (Rock and Mack 1998) and related phenomena indicate how crucial is attention to allow awareness of specific components of the outside world. Even though there is the possibility that there are other routes for stimulus activity in the brain to reach consciousness, it is accepted by most researchers in the field that attention still plays a very big role.
We thus have the situation that understanding attention better would be likely to help us get a firmer handle on consciousness. There may still be phenomena to be discovered in the future, or certain paradigms to be analysed further, which indicate that consciousness can arise without attention. However until these are properly validated we are left without any strong evidence or any clear mechanism as to consciousness arising outside attention. The large majority of cases where stimulus awareness occurs is from attention being focussed on the stimulus; only in very special cases (such as after 10 hours of training or suitable change blindness experiences) could consciousness possibly arise outside attention. We will thus concentrate in this article on how the CODAM model can enlarge our understanding of attention, and thence how CODAM can fill in some of the dynamics of the creation of consciousness.
The CODAM model is one proposal as to what this extra complexity might be in the attention control circuitry so as to create the 'owner' or 'I' of phenomenological experience: CODAM includes a copy of the attention movement control signal. This copy signal, it is proposed, indicates what we are about to receive by attending to a stimulus to bring it into consciousness.
Attention as a Controller
There is considerable and developing knowledge about attention. The paradigms being used to explore attention indicate there to be a complex interplay between numbers of brain modules across swathes of the brain as well as a complex temporal dynamics involved in this interplay. There are also further aspects of attention: its split focus character, its involvement in the initial stages of training and the general control of automated stimulus-response patterns, and the presence of buffer working memory sites that may be the sites of the emergence of the contents of consciousness (as proposed by many psychologists over the years). These complex features arise both for the exogenous and endogenous forms of attention. Recent ERP (denoting 'event related potential', obtained by averaging over numerous repeats of EEG measurements from subject's scalps) results indicate that there is also complexity in the interaction of the two forms of attention. For example the N2pc, an ERP indicator of the focus of attention, is observed to switch from the side of the brain processing a salient distracter to that processing a top-down target (Hickey et al, 2006).
There are other features that have to be considered, such as the two-stage character of attention processing now evident in the attentional blink (Vogel et al, 1998). This phenomenon indicates that there is considerable processing at low level in cortex that is outside consciousness. However this non-conscious activity can still exert its effects, as many experiments using priming by subliminal stimuli show. Thus we have to consider in what way this non-conscious activity is brought into consciousness by attention. It has been suggested by many that this 'breakthrough to awareness' occurs by focussing attention on below-consciousness activity so the activity becomes suitably amplified into awareness. There is also the need to consider how exogenous activity can break though into attention and hence to awareness, if possible as part of the same overall model.
We will thus proceed to consider how the inner workings of attention could be so constructed to enable it to explain the majority of these phenomena, and in particular as to how non-conscious stimuli could be brought by attention into consciousness, with especially the component of phenomenal experience.
The surprising feature is that in the process the extra of the CODAM model, that of the corollary discharge, can lead to an explanation of crucial aspects of 'phenomenological experience'.
To analyse attention we will take an approach that regards attention as a high level controller. The attention control system will thus consist of a higher-order control system in the brain (based in prefrontal and parietal cortices), together with lower level cortices whose activity (mainly stimulus or response-driven) is under the control of the higher order attention control system. We can take inspiration from experimental data that lead us to posit an attention movement control signal (generated by the attention control system itself, in higher cortices) that sends an amplificatory/inhibitory signal to lower level cortical regions carrying activity representing a stimulus to be attended; thus the attended stimulus activity is to be amplified or distracter stimulus activity is to be inhibited. Such a ballistic model of attention control - like shooting at a target so that once the bullet has been fired from the gun there is no further control over it - is along the lines of the "biased competition" model of Desimone and Duncan (1995), and has been tested by various groups, including that of the Posner benefit paradigm (Taylor and Rogers, 2002). In this paradigm a subject is cued either to the left or right of a stimulus screen. A target is subsequently shown on the screen. If the cue had been valid, so on the same side as the target, the reaction time for response to the target is shorter than in the invalid case (when the cue and target are on opposite sides). It is the dependence of the reduction in response time for the valid as compared to the invalid cue, as a function of the cue-target time difference, which has been successfully modelled (Taylor & Rogers, 2002) by a ballistic attention model. However the ballistic control model only begins to touch the possible richness of modern control theory, with its state estimators (or observers), error monitors and other related structures (such as delay components, etc.).
The CODAM approach has been developed in Taylor (2000, 2003, 2005, 2006a, 2006b), using the basic idea that the state estimator for attention is that of the attended state of the world, not the total state (as observed by the retina or as coded in lower level cortices). This attended state estimator has been proposed to act in the manner of a buffer working memory (slave) system, as proposed in the component model of working memory of Baddeley and Hitch (Baddeley, 1986); there is now strong evidence for the existence of this component in various modalities from brain imaging.
The structure of the resulting model shows a clear distinction from that for motor control in the brain (as developed by many researchers). The main distinction is that motor control models appeal to estimates of the total 'plant' (the body) which is being controlled. However attention control acts in quite a different manner in the brain (as noted in the previous paragraph): the estimated state is only of the attended stimulus and not of the total environment. Attention thus acts as filter, to concentrate processing resources on a particular salient stimulus and remove distracters (of which there are usually many in a complex environment). The attended stimulus, held for a period on the attended state estimator buffer (of working memory) can then be processed by further higher level mechanisms as might be involved, for example in thinking, reasoning, comparison, imagining and so on.
The filter process of attention thus gives attention a completely different control structure as compared to standard engineering control models as well as to that employed by the brain in motor control (where as many of the states of the body as is possible need to be estimated).
A valuable component of modern control theory is a forward model or predictor of the future state of the system, created early in time by an efference copy or corollary discharge of the control signal. Such a signal is well-known in motor control in the brain (Desmurget and Grafton, 2000). It has also been posited to occur and be used by the attention control circuitry of the brain (Taylor, 2003). Such a control component was conjectured to act so as to give an early boost to the buffer working memory activity arising from the neural representative activity from a specific attended stimulus attempting to broach the barrier to access to the buffer site. This mode of operation of the corollary discharge was used in a successful simulation of the attentional blink (Fragopanagos et al, 2005), as described briefly later; it employed the CODAM model, so incorporated the postulated corollary discharge signal of the attention movement control signal as an early "wake-up" call to the relevant sites to speed access to the buffer site and prevent distracter interference.
CODAM thus expands the ballistic control model of attention to a further subnetwork based on the presence of a corollary discharge (or so-called efference copy) of the attention movement control signal. There is thus the following extension of attention control models, as shown in the figure, starting with:
The Ballistic Control (Initial) model: This is composed of the following modules: (a) Input modules (early visual hierarchy in vision, for example); (b) Inverse model controller (IMC), as the generator of a feedback attention control signal to amplify the attended stimulus activity and reduce that of distracters (and acting as a saliency map, running a competitive process to attend to the most salient stimulus input); (c) Goals module, to allow for endogenous bias to the IMC for goals created from other sources in a top-down manner, or for exogenous 'breakthrough' for goals to cause attention to be directed to new and more salient inputs of greater import (but which may still generate early goal signals in FEF, so creating top-down bias for attention re-direction, following the results, for example, of Kincade et al, 2005 and Foxe & Simpson, 2002.
The Corollary Discharge Extension model (the crucial basis of CODAM) is as follows. A copy of the feedback attention movement control signal is used to generate speed-up in the access of content to report by earlier entry to the relevant buffer working memory (so gain reportability). This copy is employed to reduce effects from distracters, and to improve attention control by preventing errors in speeded response (as has been suggested to occur by use of the corollary discharge in motor control, Desmurget & Grafton, 2000). There are thus a set of further modules that are introduced into the minimal ballistic attention control model, as shown in the figure, as follows: (d) A buffer working memory site for access to report, acting as a repository of an attention tag to lower level activity, and proposed as identical to the slave sites of the distributed working memory model of Baddeley and Hitch (Baddeley, 1986). This buffer working memory activity can be considered as the attention control estimate of the attended stimulus. There is coupling of the buffer site activity to lower level attended activity, such as through amplification or by synchronisation; (e) A corollary discharge buffer, to allow the corollary discharge signal to be used to give an early prime to the working memory site (as used in a recent model of the attentional blink in Fragopanagos et al, 2005) and to activate any error signal if the goal set up earlier is not realised. The corollary discharge signal thus acts as a crucial component in prediction of forthcoming input to the buffer working memory site, so acts as a forward model; (f) An error monitor to be used to generate an early error signal from the corollary discharge signal, in comparison to the goal, so as to cause attention to be amplified to achieve its purpose, as well as to inhibit distracters from accessing the slave working memory site before reportability of the attended stimulus.
These various additional modules were used in a recent neural simulation of the attentional blink (Fragopanagos et al, 2005); they have also been used in achieving the executive function of rehearsal (Korsten et al, 2006). This latter simulation uses the error signal generated by a desired working memory signal decaying below a specific threshold, and thereby being in danger of dying away. The error signal causes attention to be re-focussed on the stimulus representation held on the buffer working memory, so preserving it for further seconds, as should occur in rehearsal.
Possible sites in the brain for each of the further modules in the Corollary Discharge extension, as well as in the Ballistic Control Initial model outlined above, have been suggested in the various references on CODAM cited earlier. To summarise, we propose the identifications of CODAM modules and brain sites (for visual attention) as (FEF = frontal eye fields, SEF = supplementary eye fields, LIP = lateral intraprietal area, SPL = superior parietal lobe, TPJ = tempero-parietal junction, IFG - inferior frontal gyrus, PFC - prefrontal cortex, DLPFC = dorsolateral prefrontal cortex, TE = anterior temporal lobe, TEO = posterior temporal lobe):
1) Input Module = Early levels of the visual hierarchy (V1, V2, V3, V4,...) There are two visual routes in this hierarchy, the M- or dorsal route for space and motion (going to V5/MT, LIP, SPL, FEF) and the P- or ventral route, going to V4, TEO, TE and IFG).
2) IMC: SPL for the dorsal route, possibly TPJ for the ventral route (although this is controversial).
3) Goal: In Prefrontal Cortex, with spatial goals coded in FEF/SEF and ventral ones in IFG (with important relations between them occurring by connections through DLPFC).
4) Sensory Buffer (WM): In various components of parietal lobes, although exact placings for feature and spatial working memory buffers are still being analysed.
5) Corollary discharge buffer: This is still only conjectured, but if it exists is expected also to be in parietal lobe.
6) Error monitor: This could be both in parietal lobe (as seen from deficit studies) as well as an important component in cingulate cortex (as observed, for example in the ERN or error-related negativity of fronto-centrally based EEG signals).
Applications of CODAM
The attentional blink requires a subject to be able to recognise a given letter, say as the first target (T1) in a rapid visual stream of stimuli presented at about 10 Hz. The subject is then required to recognise a further letter (T2) presented several lags later. The success level in recognising T2 as the lag is increased from 1 to 10 has a well established U-shape; the dip of the U occurs for a lag of about 3, or for a time gap between T1 and T2 of about 300 ms.
A detailed simulation of the attentional blink has been presented recently (Fragopanagos et al, 2005). This uses the interaction between the P3 of T1 (assumed to be created on the sensory buffer) and the N2 of T2 (assumed created from an efference copy of the attention movement control signal). The N2 is itself observed to be complex in its sources in the brain (Hopf et al, 2000; Ioannides & Taylor, 2003).
Recent preliminary evidence for the presence of both the appropriate corollary discharge signal and its protective mode of action was published by (Sergent et al, 2005). They investigated the brain patterns of activity by EEG using the Attentional Blink, and found that on those trials when the subject had failed to be aware of the second target T2 the P3 activity for the first target was shortened; this was consistent with inhibition arising from the N2 ERP of T2. This was as predicted from the circuitry of CODAM used in (Fragopanagos et al, 2005). There are other interpretations of the P3 attention blink data, and also other signals involved that may be important in the processing (such as 20 or 40 Hz synchronous activity). However the latter are expected to be part of the overall signal processing that can easily fit into CODAM; it is not specified in CODAM how binding of various features may occur, but this depends on experimental data (although timing is crucial as to when the synchronisation is observed to commence). Other interpretations can be searched for in the observed inhibition of the P3 by the N2 of the second target in the attentional blink results reported in (Sergent et al, 2005); the interpretation given here is that the N2 of the second target is acting as the corollary discharge, a crucial piece of experimental support for CODAM. Further experimentation is needed to give fuller verification of this hypothesis. However the important feature of CODAM is that it provides an approach to consciousness which is closely related to experiments in attention and its thereby testable (or possibly CODAM is modifiable to fit closer to the data).
The result of an extension of the original model of (Fragopanagos et al, 2005) by addition of inhibition from the corollary discharge buffer of CODAM to other nodes on the sensory buffer was simulated for various levels of the inhibitory connection strengths. It was shown that there was a progressive change of the activity at various lags as the inhibition was increased; this was particularly clear for the P3 of T1. This fitted well with data arising from the experiment of (Sergent et al, 2005), who observed such inhibitory interaction of the N2 of T2 with the P3 of T1, as described above.
An application of CODAM was made in (Korsten et al, 2006) to the rehearsal phase of executive control. Here the goal was set up in prefrontal cortex to preserve the level of stimulus activity on a buffer site above a critical value. If there was an error generated from the monitor, then attention was dragged back to the buffer site and amplification of the relevant activity at that site achieved (as should occur in rehearsal). Such rehearsal was applied to a paradigm (Pessoa et al, 2002) with satisfactory agreement with experiment, although many more paradigms using rehearsal (such as the N-back task) need to be simulated by a similar mechanism.
CODAM and Consciousness
The resulting CODAM model, especially in the extension to the Corollary Discharge version from the Ballistic Control Initial model, can be seen to provide an interesting mechanism for the creation of conscious experience. The content of experience is to be associated with the activation of the buffer working memory site by the incoming attention-amplified activity representing the attended stimulus (as suggested by numerous researchers over the last several decades). However the earlier corollary discharge signal is now interpreted in two ways: as a signal of the ownership of the about-to-be-experienced content, as well as a signal of guarantee of the content being that to which attention had been turned. The first of these aspects may be interpreted as the "pre-reflective" or 'inner' self of Western phenomenology (Zahavi, 1999). The second aspect is related to the important "immunity to error through misidentification of the first person pronoun", a term coined by Shoemaker (1968) and deemed of crucial importance in specifying the "I". This latter property is provided by the efference copy of the attention movement control signal guaranteeing that the content of awareness will be what the pre-reflective self is designed to "own". This ownership experience is achieved by suitable inhibition by the corollary discharge signal of any distracters on the buffer site attempting to sneak in and destroy the accuracy of the attention control ability. The content-free character of the ownership signal is expected from the lack of binding of the corollary discharge signal to lower level cortices (thought to be crucial for content to be given form as the experience of that content).
The possibility of explaining the phenomenon of 'Immunity to Error through Misidentification of the First Person Pronoun' is of crucial importance to any detailed model claiming it can begin to explain the creation of consciousness. This is clearly absent in other models of consciousness, but arises almost directly from the interpretation of the corollary discharge signal both as providing a precursor signal of 'ownership' of the about-to-be-experienced content as well as acting as a 'sentry at the gate' of access to the buffer working memory site. The gatekeeper allows entry only to that signal representation which agrees with the predicted sensory consequences provided by the corollary discharge signal itself. This is why the models of the CODAM type, using the corollary discharge signal in standard control fashion (but now with the difference mentioned earlier that attention filtering brings about an attended state estimation model, and not an estimate of the whole plant) can all be considered as 'Attention Copy Experience' models.
Much further testing of the proposed attention copy mechanism for providing the inner self of consciousness is needed both to validate the model as well as to make clearer the manner in which the flow of information occurs. If the model stands up to new data, however, then it provides an important function for attention: that of speeding up the attainment of consciousness by incoming stimuli gaining access to their relevant buffer site, as well as providing a rapid signal of possible errors.
CODAM and the Self
The manner in which the pre-reflective self may be created by the CODAM attention control architecture has been explored briefly so far, and it has been shown how this 'ownership' component of the self could arise through use of a copy of the attention movement control signal. We now turn to the manner in which the reflective self could be related to this inner self. The reflective self may be defined as that component of self which is used externally to identify ourselves, as in estimates of the need for a shave (an external estimate given by viewing a mirror or running one's hand along one's chin) or in the level of self-esteem we possess and if we are not too worried about our external image (as in periods of self-assessment or meditation). It is these external components of self - the reflective self of mirror and internal reflection - which we must attempt to relate to that component of our minds that is the inner self or 'I'.
There have been numerous recent studies of how the brain codes for various aspects of the reflective self. In their recent meta-analysis of 27 PET and fMRI experiments on self-referential processing, Northoff et al, (2006, p. 440) concluded that 'Self-referential processing in the CMS [cortical midline structures] constitutes the core of our self and is critical for elaborating experiential feelings of self, uniting several distinct concepts evident in current neuroscience.' The authors could find no functional specialisation amongst the three main clusters of activity they detected in the mid-line active brain regions of the experiments. It is unclear how far there was included any pre-reflective self activity (other than that occurring in all such experiments as part of the subjects' inner selves); all the paradigms studied involved mainly comparison of assessing traits of the self versus those of others or of determining if oneself was the agent of an action versus another. Thus we can conclude that one's reflective self is coded in a distributed manner across various midline structures in the various components of cingulate, in posterior parietal, orbitofrontal, medioprefrontal and retrosplenial cortices.
The reflective self can fit into CODAM in terms of special parts of episodic memories, such as occur in the Relational Mind approach to consciousness (Taylor, 1999), and would be brought to conscious experience by attention in the manner described earlier in the more refined CODAM approach. Through attention applied to these memory representations, they are expected to be able to create related activations on a suitable 'reflective self' working memory buffer, which then allows usage of the relevant material for use at a higher level, such as in thinking about oneself. The control approach provide by CODAM, already explored briefly earlier for rehearsal, gives an initial indication of how such thinking could be simulated in CODAM. Guidance would thereby arise from further higher level goals, so keeping the thinking along a specific track, such as to seek out one's weaknesses.
CODAM for DD and Schizophrenia
DD denotes depersonalisation-derealization. This is quite a common state which can be experienced through sleep deprivation, travel to unfamiliar places or by marijuana intoxication. It consists of a subject feeling empty inside, so that a subject could say 'I do not exist' or even 'I am dead' when in a DD state. This can be interpreted most simply as the loss of the subject's pre-reflective self. This is partly supported by experimental data from (Papageorgiou et al, 2002), who observed DD subjects by EEG measurements while the subjects performed various attention-based tasks. It was found that the DD subjects, compared with controls, had a lower amplitude of the P300 wave in parietal cortices; this wave has been taken by many neuroscientists to indicate access of relevant material to a buffer working memory site and hence to consciousness. This paradigm could be used to proceed further back in time to investigate if there was some defect in the earlier N2 wave (at about 200 ms after stimulus onset) to cause some defect in the P300 creation. Such a result would allow for important support for CODAM as well as provide a CODAM-based explanation of DD. In this there would be reduced activity of the corollary discharge, so leading to a less active inner self.
A second mental problem very relevant to CODAM arises in schizophrenia. This involves a spectrum of disorders, but in particular in its development in adolescence there is an alteration of experience of great interest: young people (usually young men) have an increased remoteness from their environment, as if they are too wrapped up in their own sensations and their own inner life. This has been termed 'hyperreflexivity', in which the subject is too involved with their own thoughts and cannot seem to control them so as to handle the nature of reality around them. It is as if their pre-reflective self has become exaggerated, so they cannot have the proper balance between the inner self and the content of consciousness. It is exactly this balance which, when lost, would present the symptoms being experienced by adolescent schizophrenics. This is the area of relevance to CODAM, since increasing the strength of the attention copy signal by one means or another would be expected to lead to such hyperreflexivity. It is to be seen if this suggestion is correct, and could be investigated by careful analysis through brain imaging of brain activity especially in schizophrenic adolescents. At a behavioural level much analysis has been given as to how schizophrenia can be seen as a breakdown of the inner self (Cermolacce et al, 2007 and references therein). There is additional data on the comparison of the size of the attentional blink between normal controls and schizophrenic patients (Wynn et al, 2006): the blink is observed to be larger for the latter as compared to the former group. Such an effect is consistent with the CODAM model so modified as have a larger N2 arising from the corollary discharge, as would correspond to an augmented pre-reflective self. Such an effect has already been simulated in the CODAM model (Fragopanagos & Taylor, 2007).
The above analysis of attention and its possible relation to consciousness has been based on attention as a control system, and the extension of that control system beyond ballistic control into the employment of a corollary discharge or copy signal of the attention movement control signal. This has led to the CODAM (COrollary Discharge of Attention Movement) generic models of consciousness, in which the pre-reflective self arises from 'experience' of the attention copy signal at an early stage in the processing. The CODAM models are a natural extension of the more primitive ballistic model of attention, and are possibly sophisticated enough to provide the necessary mechanism mentioned earlier as needed, inside attention, to generate important components of consciousness itself. The CODAM signal - the attention copy signal - occurs before the attention amplification of the stimulus activation has been achieved, itself leading to the experience of the content of consciousness. Thus a possible temporal flow of consciousness is predicted by the CODAM-type models, which need to be analysed much further.
- Baddeley AD (1986) Working Memory. Oxford: Oxford University Press.
- Block N (1995) On a Confusion about a Function of Consciousness. The Behavioural and Brain Sciences 18(2)
- Cermolacce M, Naudin J & Parnas J (2007) The "minimal self" in psychopathology: Re-examining the self-disorder in schizophrenia. Consciousness and Cognition 16:703-714
- Desimone R & Duncan J (1995) Neural mechanisms of Selective Visual Attention. Annual Reviews of Neuroscience 18:193-222
- Desmurget M & Grafton S (2000) Forward modelling allows feedback control for fast reaching movements. Trends in Cognitive Sciences 4:423-431
- Foxe JJ & Simpson GV (2002) Flow of activation form V1 to frontal cortex in humans. Experimental Brain Research 142:139-150
- Fragopanagos N & Taylor JG (2007) Resolving some confusions over attention and consciousness. Neural Netowrks 20:993-1003
- Fragopanagos N, Kockelkoren S & Taylor JG (2005) Brain Res Cog Brain Res 24:568-586
- Hickey C, McDonald JJ & Theeuwes J (2006) Electrophysiological Evidence of the Capture of Visual Attention. Journal of Cognitive Neuroscience 18(4):604-613
- Hopf J-M, Luck SJ, Girelli M, Hagner T, Mangun GR, Scheich H & Heinze H-J (2000) Neural Sources of Focused Attention. Cerebral Cortex 10:1233-1241
- Ioannides AA & Taylor JG (2003) Testing models of attention with MEG. Proc Int J Conf Neural Networks 1:287-291
- Kincade JM, Abrams RA, Astafiev SV, Shulman GL & Corbetta M (2005) An Event-Related Functional Magnetic Resonance Imaging Study of Voluntary and Stimulus-Driven Orienting of Attention. Journal of Neuroscience 25(1`8):4593-4604
- Korsten NJH, Fragopanagos N, Hartley M, Taylor N & Taylor JG (2006) Attention as a controller. Neural Networks 19(9):1408-1421
- Mack A & Rock I (1998) Inattentional Blindness. Cambridge Mass: MIT Press
- Northoff G, Heinzel A, de Greck M, Bermpohl F, Dobrowolny H, Panksepp J (2006): Self-referential processing in our brain – A meta-analysis of imaging on the self. Neuroimage 31(1):440-457
- Papageorgiou C, Venouras E, Uzunoglu N, Rabavilas A & Stefanis C (2002) Changes of P300 Elicited during a Working Memory Test in Individuals with Depersonalisation-Derealisation Experiences. Neuropsychobiology 46:70-75
- Pessoa L, Gutierrez E, Bandettini P & Ungerleider L (2002) Neural correlates of visual working memory: fMRI amplitude predicts task performance. Neuron 35:975-987
- Sergent C, Baillet S & Dehaene S (2005) Timing of the brain events underlying access to consciousness during the attentional blink. Nature Neuroscience Sep 11, 2005
- Shoemaker S (1968) Self-reference and self-awareness. The Journal of Philosophy 65:555-567
- Taylor JG (2006a) On the neurodynamics of the creation of consciousness. Cognitive Neurodynamics 1: (in press)
- Taylor JG (2006b) The Mind: A User's Manual. London: Wiley & Sons
- Taylor JG (2005) Mind and consciousness: Towards a final answer? Physics of Life Reviews 2:1-45
- Taylor JG (2003) Paying attention to consciousness. Prog in Neurobiology 71:305-335
- Taylor JG (2000) Attentional movement: the control basis for consciousness. Soc Neurosci Abstr 26:2231 (#839.3)
- Taylor JG (1999) The Race for Consciousness. Cambridge Mass: MIT Press
- Taylor JG & Rogers M (2002) A control model of the movement of attention. Neural Networks 15:309-326
- Vogel EK, Luck SJ & Shapiro KL (1998) Electrophsyiological evidence for a postperceptual locus of suppression during the attentional blink. J Exp Psychol Hum Percept Perform 24:1656-1674
- Wynn JK, Breitmeyer B, Nuechterlein KH & Green MF (2006) Exploring the short term store in schizophrenia using the attentional blink. Journal of Psychiatric Research 40:599-605
- Zahavi D (1999) Self-awareness and Alterity. Evanston Ill: Northwestern University Press
- Jan A. Sanders (2006) Averaging. Scholarpedia, 1(11):1760.
- Valentino Braitenberg (2007) Brain. Scholarpedia, 2(11):2918.
- Olaf Sporns (2007) Complexity. Scholarpedia, 2(10):1623.
- James Meiss (2007) Dynamical systems. Scholarpedia, 2(2):1629.
- Paul L. Nunez and Ramesh Srinivasan (2007) Electroencephalogram. Scholarpedia, 2(2):1348.
- William D. Penny and Karl J. Friston (2007) Functional imaging. Scholarpedia, 2(5):1478.
- Seiji Ogawa and Yul-Wan Sung (2007) Functional magnetic resonance imaging. Scholarpedia, 2(10):3105.
- Daniel J. Simons (2007) Inattentional blindness. Scholarpedia, 2(5):3244.
- Anil K Seth (2007) Models of consciousness. Scholarpedia, 2(1):1328.
- Peter Jonas and Gyorgy Buzsaki (2007) Neural inhibition. Scholarpedia, 2(9):3286.
- John Dowling (2007) Retina. Scholarpedia, 2(12):3487.
- Ernst Niebur (2007) Saliency map. Scholarpedia, 2(8):2675.
- Julian Lim and David F. Dinges (2007) Sleep deprivation. Scholarpedia, 2(8):2433.
- Arkady Pikovsky and Michael Rosenblum (2007) Synchronization. Scholarpedia, 2(12):1459.