While we are all happy with the role of attention in the learning process, I am afraid the same cannot be said for the construct of consciousness or awareness and its role during this process. Here is one of my favorite statements: “Con-sciousness as an object of intellectual curiosity is the philosopher’s joy and the scientist’s nightmare” (Tulving 1993: 283). Tulving certainly knew what he was saying, given that the “multifaceted nature of the construct of ‘awareness’
makes it undoubtedly one of the slipperiest to operationalize and measure in both second language acquisition (SLA) and non-SLA fields such as cognitive psychology, cognitive science, and neuroscience” (Leow, Johnson, & Zárate-Sández, 2011: 61). In addition, the conf lation of the terms “consciousness”
and “awareness,” which you probably already noticed in previous chapters, is quite remarkable—for example, “any evidence that perception is not neces-sarily accompanied by an awareness of perceiving attracts attention because it challenges the idea that perception implies consciousness” (Merikle, Smilek, &
Eastwood, 2001: 116). In a similar vein, Schachter (1989) used the term “con-sciousness” interchangeably with “phenomenal awareness” while referring to Dimond’s (1976) definition of consciousness as “the running span of subjective experience” (p. 377). More recently, Friedenberg (2013) not only underscored the slipperiness of the construct of consciousness but also immediately conf lated consciousness with awareness: “Consciousness is perhaps one of the greatest mys-teries in the universe. How are we aware of ourselves?” (p. 3) .
This chapter reports on several varieties of consciousness and revisits the role of consciousness in non-SLA information processing models with respect to its association with the metaphor of the human being as limited capacity processors, a limited capacity information selection system, and the concept of a limited capacity central executive. The global neuronal workspace theory is
also discussed, with its key feature of perceiving the mind as modular. This is premised on the hypothesis that automatic or unconscious cognitive processing depends on multiple processors or modules. The relevance of these models to the L2 is discussed, followed by definitions of what comprises the construct of consciousness in non-SLA fields and the way this construct has been operational-ized and measured in empirical research in these fields. But first, here are some musings on the construct of consciousness.
Some Musings on Consciousness
Before I go into a concise report of the non-SLA theoretical underpinnings for the role of consciousness (and if you also noticed, I use the term “conscious-ness” in non-SLA fields but the term “aware“conscious-ness” in SLA), here are a couple of questions to ponder. Where does consciousness reside? If one were to respond
“in the brain,” then does consciousness have one home address or more? Does it skip from one neural network to another? How do we know someone is con-scious of some data in the input? Is someone’s concon-sciousness the same as another person’s, say, during a similar task or interaction? Indeed, what is consciousness ? I am sure you are aware that the answers are not easy, but later in the chapter I am going to provide you with some tentative definitions as we take a quick look at empirical studies purporting to address implicit learning in the non-SLA fields.
However, there may be a few facts with which we can all agree, and later, I shall summarize others gleaned from the attentional theories in non-SLA fields. First of all, consciousness is a subjective experience, and we all experience events differently, which appears to indicate that there may be levels of consciousness.
Friedenberg (2013) put it in an interesting way: “Although we may be able to better understand what it is like to be another human hearing a Beethoven sym-phony, smelling a rose or seeing a Monet painting, we can never be sure we are actually having the same exact experience ourselves” (p. 3). Friedenberg also pointed out the mismatch between science, being objective in its scope, and a subjective phenomenon like consciousness. Scientists can talk about neurons firing more rapidly in specific parts of the brain as blood courses through the brain when compared to other regions, but they cannot tell us much about what it is like to be that person whose brain is being analyzed or, more specifically, the individual’s subjective quality of experience. Indeed, from an experimental perspective, does one’s consciousness (or awareness) rise as they know they are about to take part in an experiment?
Varieties of Consciousness
There may be varieties of consciousness if one were to accept that conscious experience can be linked to heightened firing of neurons in specific parts of the brain (does this mean that consciousness resides in several homes?). In this
perspective, we can differentiate different types or categories of consciousness;
for example, brain activity is differential between, say, being awake or being in different stages of sleeping. Levels of consciousness are also assumed, which can range from being in a conscious state to being in a comatose state to being in a coma (is any level pertinent to some of our students?). Other researchers (e.g., Kihlstrom, 1984) differentiate levels that include conscious, preconscious, and unconscious (cf. Dehaene, Changeux, Naccache, Sackur, & Sergent, 2006, who discussed processing levels that include conscious, preconscious, and subliminal), which, overall, appears to indicate minimally a three-level home (cf. Block, 1995, for a four-fold classification of conscious states). We can also identify two aspects of consciousness, namely, the state of consciousness, as in, for example, being awake or in a coma, and the contents of consciousness, that is, being conscious or aware of a specific item in the input. While these two are not mutually exclusive (Dehaene & Changeax, 2005), we are focusing on the latter aspect in this book.
Okay, now that we are all conscious and aware of the slipperiness of this construct (think of trying to get hold of an eel), let us discuss theoretical under-pinnings in the non-SLA fields. Just keep in mind, once again, as you read the information below that most of this discussion is provided within the visual attention field, so if we were to throw in natural language processing with all its complexities, the picture does become quite nebulous, doesn’t it?
Theoretical Non-SLA Models of Attention Revisited
If you recall in Chapter 3 , we discussed the theoretical models of attention from an information processing perspective, which views humans as limited capacity processors of information contained in the input. We also noted that the pro-cess of attention was usually addressed via visual attention. We also discussed the notion of several stores postulated to exist along the learning process, from sensory registers to short-term storage of information to long-term storage of said information, with the ensuing discussion of where we should install a filter, leading to the removal of such a filter. We also discussed attention from a neuro-scientist perspective, in which this construct or process is viewed as being more fine-grained, occurring in different parts of the brain (a modular perspective), instead of coarse-grained (e.g., a metaphor). Like the rationale for the selection of attentional models in non-SLA fields, let us revisit some of these attentional theo-ries, to see what role is assigned to consciousness or awareness, and also other the-ories that have played a role in some SLA theoretical underpinnings and research.
The Role of Consciousness in Non-SLA Information Processing Models
Schmidt (1990) provided quite a concise summary of the notion of conscious-ness in these information processing theories in psychology. He reported that it is usually associated with this metaphor of the human being as limited capacity
processors. Within this metaphor, he identified the notion of consciousness in three ways. The first is associated with the contents of a limited capacity memory system. As you will recall, information processing models typically posit a series of storage structures to account for input processing, ranging from a sensory register, in which pre-attentive and usually unconscious processes are employed to select information from the input, to a short-term memory store (working memory) to a final long-term memory store. Consciousness is usually identified with short-term memory that in turn is often conf lated with the construct of consciousness and focal awareness (Kihlstrom, 1984), and processing in short-term memory is regarded as essential for long-short-term memory. In other words, information held in short-term memory or working memory, unless further processed, will most likely disappear from memory (e.g., Baars, 1988; Cowan, 1999; Kihlstrom, 1984; Logan, 1988; Nissen & Bullemer, 1987; Posner, 1992).
Viewed from this perspective, one postulation may be that the role of conscious-ness is crucial for learning to take place. Keep in mind, though, that it may not be as simplistic as viewing the learning of some feature of the L2 input as either with or without awareness, given the many variables that can be associated with the learning process (e.g., type of linguistic item, amount of prior knowledge, motivation or interest, language proficiency, social setting, etc.).
The second association, according to Schmidt, is with a limited capacity infor-mation selection system. Consciousness in other inforinfor-mation processing models, especially the early filter models discussed in Chapter 3 (e.g., Norman, 1968), is linked to attention viewed as a control process that transforms information (e.g., in the detection storage structure) into focal awareness or as a limited resource, as in the attentional resource models (e.g., Wickens, 1984), where the notion of effort is embedded in the simultaneous performance of two or more tasks. The notion of attention as a resource is also viewed as a distinction between two types of process-ing, namely, controlled versus automatic processprocess-ing, where controlled processing is assumed to be under conscious control (Posner & Snyder, 1975; Shiffrin & Schnei-der, 1977) and typically associated with new or novel information, while automatic processing requires little, if any, mental effort to process the incoming information.
The third association has to do with the concept of a limited capacity central executive. In this concept, consciousness is viewed as an internal programmer, executive control center, or a supervisory attentional system to address some sort of planning or making of critical decisions (e.g., Norman & Shallice, 1986). The limited capacity exists in the ability to coordinate mental activity during input processing.
There are several other theories or models in non-SLA fields that posit some role for consciousness, but more importantly, these theories or models do allow for the integration of selective attention, working memory, and cognitive con-trol together with consciousness. I am going to select only those that have in some way provided the theoretical foundation for a few current SLA theoretical underpinnings and brief ly mention a few others that underscore the care we need to exert when addressing these theoretical underpinnings.
The Global Neuronal Workspace
Before beginning, it is important to draw your attention to and make you aware, once again , of the very important issue of what type of data was gathered to sup-port postulations as to the role of consciousness, namely, perceptual data that were mainly visual and auditory.
Probably the most known non-SLA theory of consciousness is Baars’ (1988) cognitive theory of consciousness, which was later developed further by other researchers that include Dehaene and colleagues (e.g., Dehaene & Naccache, 2001; Dehaene, Sergent, & Changeux, 2003). Premised on empirical findings on (1) the depth of unconscious processing, (2) attention as a prerequisite of consciousness, and (3) the necessity of consciousness for some integrative mental operations, they postulated the “hypothesis of a global neuronal workspace.”
One key feature of the global neuronal workspace theory is the perception of the mind as modular (e.g., Fodor, 1983). This is premised on the hypothesis that automatic or unconscious cognitive processing depends on multiple processors or modules (Baars, 1988; Fodor, 1983; Shallice, 1988), with at least two func-tional and neurobiological definitions dependent upon field of study. Dehaene and Naccache (2001) pointed out that in cognitive psychology, modules are char-acterized by their information encapsulation, domain specificity, and automatic processing, while in neuroscience specialized neural circuits responsible for pro-cessing specific types of input have been identified (e.g., via brain imaging, neu-ropsychological dissociations, and cell recording) at several spatial scales, ranging from orientation-selective cortical columns to face-selective areas. As informa-tion enters, control is distributed across these processors and what is known as a global workspace or central information exchange. Indeed, Baars (1988) neatly described this global workspace as a broadcasting station that receives informa-tion from different sources and shares this informainforma-tion with its listeners.
Dehaene and Naccache (2001) proposed that “a given process, involving sev-eral mental operations, can proceed unconsciously only if a set of adequately interconnected modular systems is available to perform each of the required operations” (p. 12). This hypothesis implies that multiple unconscious operations can proceed in parallel as long as they do not simultaneously appeal to the same modular systems in contradictory ways (cf. Wickens’ (1989) notion of parallel versus serial processing). In addition, unconscious processing may occur at both low-level, that is, computationally simple, operations or high-level operations, but this kind of processing needs to be associated with “functional neural path-ways either established by evolution, laid down during development, or automa-tized by learning” (p. 13).
On the other hand, consciousness depends upon access to this global work-space; conscious experience is informative, and adaptive processes in the nervous system are triggered by conscious events. According to Baars (1988), the learn-ing process begins with the realization that there is somethlearn-ing to be learned
and undergoes several stages that set the foundation for understanding the new information. Once this foundation for understanding is established (internal-ized?), the new information fades out of consciousness and becomes part of the unconscious foundation subsequently employed for interpreting new informa-tion (think again of activainforma-tion of prior knowledge). In other words, new infor-mation does not appear to be candidates for implicit or unconscious processing if it is assumed that some connection to previous information is necessary for this type of processing and that some foundation has already been laid down, perhaps via multiple exposures over some period of time, to support such implicit processing. Think of this process as the establishment of knowledge that is sub-sequently activated to facilitate and process incoming information (or the role of prior knowledge facilitating comprehension, quick retrieval of an associated linguistic form or structure, and learning). Dependent upon the use of prior knowledge, depth of processing or how human beings process incoming infor-mation may be affected in the early stages of the learning process.
To visualize the workings of this concept of a global workspace, let us take a look at a model of global neuronal workspace theory in relation to the well-researched attentional blink (e.g., Dehaene et al., 2003). In this paradigm, par-ticipants are exposed to two successive stimuli presented at two different time intervals. If the interval between the presentations of the two stimuli is short, participants’ ability to report the second stimulus decreases (as in if their atten-tion “blinks”). To situate this finding within the model premised on a hierarchi-cal nature of cortihierarchi-cal organization, Dehaene et al. (2003) postulated that when the first stimulus was presented, the network created a global state in which the initial stimulus is represented at all levels of the hierarchy. Due to its recurrent connections, this information may remain for a short period. If the second stim-ulus is introduced shortly after the first stimstim-ulus, it faces top-down competition from the lingering representation of the first stimulus and cannot be processed effectively (sounds like Wickens’ notion of serial processing, doesn’t it, or per-haps to be more up-to-date, multi-tasking?). Interestingly, while the top-down inf luences in this model derive from the lingering activation of the previous stimulus, Dehaene et al. have observed that other models have identified this top-down inf luence as representing the focus of attention (e.g., Corchs & Deco, 2002; Spratling & Johnson, 2004), items in working memory (e.g., O’Reilly, 2003; O’Reilly, Braver, & Cohen, 1999), and task demands (e.g., Cohen, Aston-Jones, & Gilzenrat, 2004).
Neuroscience and Connectionist Modeling
Maia and Cleeremans (2005) examined the construct of consciousness from both cognitive neuroscience and connectionist modeling, underscoring the view that in these areas selective attention, working memory, cognitive control, and consciousness involve competition between widely distributed representations,
which are biased by top-down processes found notably in the prefrontal cortex in the brain.
According to Maia and Cleeremans, these models, including the global work-space theory, embody the fundamental principle used in connectionist models decades ago, namely global constraint satisfaction. They listed the mechanisms hypothesized to be associated with consciousness, provided below (p. 397):
1. Active representation
Active neuronal firing is necessary (but probably not sufficient) for consciousness.
2. Global competition biased by top-down modulation
Global competition between representations leads to consciousness. The winning neuronal coalition determines both conscious phenomenal expe-riences and global accessibility. Active representations maintained by the prefrontal cortex (PFC) are important sources of biases for this competition (Read: higher levels of consciousness based on neuronal firing rate).
3. Global constraint satisfaction
Global competition implements global constraint satisfaction. Con-scious experience can be viewed as the result of a large-scale application of the brain’s knowledge to the current situation (Read: the role of prior knowledge).
4. Reentrant processing
Recurrent connections are essential to implement global constraint sat-isfaction. They allow more global interpretations in higher-level areas to inf luence processing in lower-level areas (which tend to work more like localized feature detection) (Read: the relationship between frequency of input and activation of prior knowledge).
5. Meta-representation
Higher levels of human consciousness and cognition, such as the ability to think about one’s thoughts, may depend on the creation of representa-tions that are then fed back to the same constraint satisfaction network as input (Read metacognition, “thinking about thinking,” or monitoring our own performance).
Like most connectionist models, two types of representations are identified (cf. Maia & Cleeremans, 2005). On one hand, there is the long-term knowledge that is embedded or latent in the weights of the connections between units.
These representations can inf luence behavior indirectly by eliciting specific fir-ing patterns of neurons over groups of units, and are not directly accessible. This is the type of unconscious prior knowledge that we employ to make sense of most of what we are exposed to. When some piece of information is difficult to process, the usual culprits may be a simple lack of prior knowledge, an activation
of inappropriate knowledge, or the inability to make the appropriate connection.
On the other hand, there are representations that are more transient and active in the form of firing patterns, and as such, according to Maia and Cleeremans, conscious representations must depend upon these active representations. The hypothesis, then, is that only the outputs of computations in the brain may be potentially conscious, while the mechanisms of the computations themselves are unconscious.
To grasp the concepts of global competition biased by top-down modula-tion, let us imagine a war of neurons taking place in the brain as information is distributed across several areas of this wonderfully and architecturally designed part of our upper body. One area of neurons is heavily armed and logically
To grasp the concepts of global competition biased by top-down modula-tion, let us imagine a war of neurons taking place in the brain as information is distributed across several areas of this wonderfully and architecturally designed part of our upper body. One area of neurons is heavily armed and logically