For a naturalist there seems no better option than some form of information integration theory of consciousness. They emphasise the physical processes and information relations that go on inside a brain and try to explain how consciousness arises from these interactions. In this essay I will focus on Lamme’s theory of recurrent processing (Lamme 2000, 2003, 2004). First we are going to need a brief description of what is meant by consciousness and what it is that Lamme and others attempt to explain. Then we will look at which aspects of this notion of consciousness Lamme tries to describe, including some evidence for his position. Having described his story and having accepted that his argument is a solid foundation on which to build, I will conclude that Lamme is successful in describing the neural correlates of the aspects of consciousness he engages with, but as he himself admits there are some gaps to attend to.
Consciousness in General
Consciousness has many apparent features. Those features all together encompass the various aspects that we think of when we consider the ‘what’ of consciousness.
Consciousness is qualitative in that there is something it is like to be having any given conscious experience. The experience of reading this essay is one that is unique to you right now, just as unique as the experience of what it is like to be a bat, to see red, to experience pain, and so on. For each conscious experience there is some qualitative thing that it is like to be having that experience (Nagel 1974).
Consciousness is coherent in that each aspect of conscious experience fits neatly into an overall picture of our world. It makes sense for there to be an experience of redness while looking at a rose, or to hear the sound of a person speaking as you watch their mouth move. It possesses a first person perspective. My conscious experience is specifically mine. It is spatially located and seems to belong only to me. It is natural to speak of my conscious experience because it is something that is happening to me. These are both aspects of what Van Gulick (2017) refers to as the phenomenal structure of consciousness.
Closely linked to both of the above is the subjective nature of consciousness. Nagel (1974) wants to say that not only is there something it is like to be a bat, but that the something, the qualia of batness can only ever be experienced by a bat. Only a bat is able to experience the world in a bat-like manner.
An extension of coherence and closely related to the first person perspective, unity, is an important aspect of conscious experience. Features of a given object seem bound together, an apple is round, red, feels a certain way in your hand, etc. All of these features are part of the apple and the apple in turn is part of my overall experience as I look at my desk. A bird flies past, it’s motion, colour, scent, and sound all form a unified picture of the bird. Combine this with the background scene and your sense of everything else around you and you find yourself at the centre of a unified conscious experience. (Brook and Ramont 2017)
Finally it possesses some aboutness. Your conscious experience is, in general terms, about the world around you. It is a kind of mental map constructed from the various sensory modalities which guides appropriate behaviour toward your environment. More specifically you can direct the aboutness of your conscious experience toward specific things. When you think about the apple, your conscious experience becomes about the apple. By changing the attentive focus of your conscious experience you change the aboutness of it. (O’Brien and Opie 2009)
It is largely toward this last aspect that we will now turn. However Lamme also has something further to say about the distinction between conscious and unconscious information.
Lamme’s Neural Correlates of Consciousness
Lamme directs much of his focus toward distinguishing between awareness and attention. He focuses largely on the visual sense modality but claims that his theory applies to all of the senses. He further distinguishes conscious and unconscious information. He makes the distinction along the lines of different types of processing that happen within the visual cortex. Those types of processing are: Feedforward, Locally Recurrent, and Globally Recurrent. (The discussion that follows relies entirely on Lamme 2000, 2003, 2004)
The Conscious-unconscious Divide
The distinction between conscious and unconsious is a relatively simple one. Much of the processing that occurs is unconscious and Lamme has two ways of distinguishing between what is and is not. The first method is to label everything that is not accessible to attention as unconscious. That is everything you cannot directly access is unconscious. For example, specific electromagnetic wavelengths entering the visual cortex are not accessible to our attention. We cannot look at an object and determine the precise wavelength that it is reflecting (without tools). The second method is to determine which information is being subjected only to feedforward processing.
Feedforward processing is the kind of processing that a neural network engages in at a very short latency to the presentation of a stimulus. The visual cortex is divided into various functional subsections. The specifics of which are not particularly important to the concept here. For present purposes it is enough to know that there are at least 5 sections, arranged roughly linearly, moving from the rear of the cortex to the front. Each side of the brain has such an arrangement which processes the information from the opposite visual field. These are labelled V1, V2, V3, V4, and V5. As the sensory input enters the system at V1, a feedforward sweep begins and information is passed successively along the chain. Entry to V1 occurs at around 40ms, and the entire feedforward sweep takes between 100 and 150ms. What is important is that the information carried in this feedforward sweep is not attendable and we are not aware of its content. Responses can be produced at this point, for example reactions to loud noises, bright lights, fast moving objects, etc. But there is no cognitive aspect to the response, no thought goes into it, no explicit decisions are made.
Two observations confirm this hypothesis. Backward masking involves the presentation of a new stimulus shortly after the first (~40ms). This causes the subject to be unaware of the original stimulus, but the information still reaches certain areas within the feedforward sweep. Backward masking does however supress recurrent processing (more on this soon). Feedforward activity is still present in anaesthetised animals. These together suggest that feedforward processing is not enough to establish conscious activity.
I will leave this discussion unconcluded with regard to conscious activity for the time being. However it will become clear in the discussion of the following section that Lamme thinks recurrent processing is essential to both the conscious-unconscious and awareness-attention distinctions. Although the recurrence is at different scales.
The Awareness-Attention Distinction
The distinction between awareness and attention is one of scale. For Lamme both of these aspects of consciousness rely on recurrent processing. For awareness that is locally recurrent, and for attention that is globally recurrent. Awareness is information that is available in the system but may not be consciously reportable. Attention is a selective mechanism, that relies on factors related to the current state of the system and its history, which is responsible for determining focus and committing to memory.
The conception of awareness here is that you are broadly aware of everything within the particular sense modality, but unless you attend to any given object or aspect of that awareness you are unable to report on it. So it seems that awareness and attention are intimately bound together. Lamme thinks, however, that they are not so much bound as attention is a selective mechanism that operates on awareness and has a preference for certain types of stimulus.
Recurrent processing in a general sense is information passed to a lower level in the relevant cortical system. So in the case of the visual cortex described above, recurrent activity is any information being passed from a higher number to a lower number. The ways in which this information is passed to lower levels and used there become quite complicated. What is occurring is that more specific information, information which has already undergone some processing, about the stimulus is being used to modulate the lower level information. This effectively results in a more refined representation of the relevant aspects of the stimulus. Locally recurrent processing occurs at a small scale, such as across a few layers of a cortical region. By continuously passing more and more refined information back to lower levels for further processing the brain is able to build up an incredibly rich and detailed picture of the stimulus. This rich picture is what accounts for phenomenal awareness. It provides the picture such that attention can be directed at it. Thus everything contained within this phenomenal awareness, everything available to be attended to is conscious information, but certain observations, discussed below, show that your ability to report on this awareness is a function of memory, not of consciousness.
So we come to our last distinction, access consciousness. When attentional selection operates to commit aspects of phenomenal awareness to memory it makes them reportable, it makes them accessible. Various experiments designed to test your sensitivity to change in your environment have been suggested to support the conclusion that you are only conscious of, or aware of, objects if you are able to report on them. What Lamme suggests instead is that what is really being tested here is your ability to recall information about a scene that is no longer present. Inattention blindness (IB) occurs when the subject is focusing on a task and is presented with some unexpected stimulus. When asked to report on the nature of the extra stimulus they are often unable to. Change blindness (CB) occurs when some aspect of a scene is changed and the change is masked, for example by interposing a blank scene for a brief period. When asked to report on what has changed the subject is again often unable to do so. To claim that what is at question here is the subject’s consciousness of the stimulus is to claim that the subject is somehow selectively conscious of aspects of the scene but not others. This, says Lamme, is the difference between phenomenal awareness and access consciousness. In both IB and CB what is really happening is that the subject is aware of but has not attentively selected the stimulus for encoding in memory. The subject was aware of the change or additional stimulus but has forgotten what it was.
Given the number of specific aspects of the scene being processed during local recurrent activity, the process of attentive selection is to acquire the most salient features. Thus attentive selection is a function of globally recurrent processing, short term memory, and long term memory. Short term memory is traces of recent stimuli which make it more likely for the neural network to respond in a certain way. Long term memory is a stable semi-permanent change in the physical layout of the network which makes it more likely to respond in a certain way even when trace activity is not present. The global aspect ensures that the features that get selected are coherent with the overall state of the subject as a whole.
So unconscious activity is that which is processed by feedforward structures only. Awareness is recurrent activity which builds the scene within a specific modality as a whole and is effectively synonymous with conscious activity. Attention is a selective mechanism which allows certain features to reach the level of access consciousness such that they can be reported on or responded to through cognitive processes.
There are some problems for Lamme’s theory however. His story is incomplete which he himself admits. That incompleteness leaves him open to some criticisms. He gives us a lot of the what but very little of the why or how.
The strongest evidence he presents for the theory is in the lesion studies which show that when there are lesions on the primary visual cortex there is significant guiding of behaviour by stimuli of which the subject is not aware, suggesting that awareness is not necessary, and supporting the role of feedforward processing in the guiding of behaviour. Also when artificial lesions are caused temporarily via external electromagnetic pulse the effect varies depending on the kind of processing that is interrupted. The largest effect is caused when stable recurrent activity is interrupted in this way, suggesting that this activity is necessary for awareness. It has been suggested that stable oscillatory activity across the specific and non-specific thalamocortical loops provide the foundations of consciousness. The evidence for this alternative is equally strong and is supported by studies showing interruptions to conscious experience where there are lesions on those loops. Lesions on the specific thalamocortical loop will effectively knock out the related modality, while lesions on the non-specific loop will knock out the background state of consciousness (Llinas 2003, Llinas et al 1998). Can it be strongly claimed that either is correct in that their particular type of stable multidirectional neural activity is the right condition for consciousness? This conflicting evidence suggests either that we are looking in the wrong direction for the answer, or that the two theories are interrelated and perhaps form parts of a greater picture. I conjecture, although it isn’t quite within the scope of this essay, that thalamocortical binding has a role to play in the global unification of consciousness across modalities.
O’Brien and Opie (2009) claim that a major problem for any process theory is that they rely only on the computational interactions in the system and as such do not have access to the internal properties and relations of the vehicles that the neuron firing patterns represent. Lamme certainly does seem to implicitly accept that the contents of the states he speaks of are representationally encoded. He speaks of various areas of the visual cortex as representing aspects of the visual field (Lamme 2003, 2004). If he accepts that neural firing patterns are representing vehicles in this way, then it must follow that either he does not think that the internal properties have any bearing, or that he does not see representation as an integral part of the functioning of the system. Either way, if one accepts that neural firing patterns constitute representing vehicles and that representation does play a role, in no way does this present an immediate objection to the theory. There are questions left open that can readily be addressed by vehicle theory, but those questions do not fundamentally alter the nature of the descriptions Lamme provides for attention, awareness, and conscious information. They seem only to effect the contents and perhaps magnitude of the conscious states.
I have described the framework within which theories attempt to describe various aspects of consciousness. I have presented one such theory which relies on three different types of processing: feedforward unconscious processing, locally recurrent conscious processing (awareness), and globally recurrent conscious processing which plays a major role in the mechanism of attentive selection. Lamme has left many questions open and presents some exciting opportunities to advance his story. I have shown how two possible objections to the recurrent processing theory of consciousness might in fact provide deeper insight into the overall picture. It is my conclusion that Lamme is successful in making clear the distinctions between awareness and attention, and conscious and unconscious processes. But that many of the gaps he leaves require some further exploration in order to be fully satisfying.
Brook, A. and Raymont, P. (2017) “The Unity of Consciousness”, The Stanford Encyclopedia of Philosophy (Summer 2017 Edition), Edward N. Zalta (ed.), forthcoming.
Lamme, V A. (2000) “Neural Mechanisms of Visual Awareness: A Linking Proposition.” Brain and Mind 1: 385-406.
Lamme, V A. (2003) “Why Visual Awareness and Attention Are Different.” Trends in Cognitive Sciences 7: 12-18.
Lamme, V A. (2004) “Separate Neural Definitions of Visual Consciousness and Visual Attention; A Case for Phenomenal Awareness.” Neural Networks 17: 861-72.
Llinas, R. (2003) “Consciousness and the Thalamocortical Loop.” International Congress Series 1250: 409-16.
Llinas, R. et al. (1998) “The Neuronal Basis for Consciousness.” Phil. Trans. R. Soc. Lond. B 353: 1841-9.
Nagel, T. (1974) “What is it like to be a bat?” Philosophical Review 83: 435-50.
O’Brien, G. and Opie, J. (2009) “Vehicles of Consciousness.” In T. Bane, A. Cleeremans, & P. Wilken, eds., The Oxford Companion to Consciousness, Oxford University Press.
Van Gulick, R. (2017) “Consciousness”, The Stanford Encyclopedia of Philosophy (Summer 2017 Edition), Edward N. Zalta (ed.), forthcoming.