Since around 1993 I have been developing and researching the predictions that fall under the scope of what I call the Cognition Hypothesis of task-based learning (Robinson, 2001a,b, 2002, 2003a,b, 2005, 2007, 2011a,b), which claims L2 tasks should be sequenced for learners on the basis of increases in their cognitive complexity alone, and not on linguistic grading. As Long (1985, 1997) has described, in a task-based syllabus pedagogic tasks are developed and sequenced to increasingly approximate the demands of real world target tasks (as opposed to conventional structural syllabuses which use tasks to deliver units of language, e.g., Ellis, 1994, Nunan, 1989). The predictions of the Cognition Hypothesis are based on my reading in areas of functional/cognitive linguistics, (e.g., Givon, 1995, 2009, Talmy, 2000, Tomasello, 2003), and in developmental psychology (e.g., Cromer, 1991, Slobin, 1993, 2004) and SLA theory. The hypothesis claims that increasing the cognitive demands of tasks contributing to their relative complexity along certain dimensions will; a) push learners to greater accuracy and complexity of L2 production in order to meet the consequently greater functional/communicative demands they place on the learner; b) promote heightened attention to and memory for input, so increasing learning from the input, and incorporation of forms made salient in the input; as well as c) longer term retention of input; and that d) performing simple to complex sequences will also lead to automaticity and efficient scheduling of the components of complex L2 task performance.
        While my work on the Cognition Hypothesis has as a primary motivating goal the development of feasible sequencing criteria for classroom tasks, it is not limited to this either in explanatory scope or in potential practical application. The Cognition Hypothesis is also important to explore for those concerned to develop equivalent forms of language tests; and for those concerned to measure gain resulting from experimental exposure accurately, by using equivalent pre and post test measures of language use, as well as for calibrating batteries of equivalently complex pre and posttests.
        In the ‘Triadic Componential Framework’ (Robinson, 2001a, b, 2003a, 2005a, 2007, 2011a,b) I have proposed for examining the implications of the Cognition Hypothesis for classroom practice and syllabus design I distinguish the cognitive demands of pedagogic tasks contributing to differences in their intrinsic complexity (e.g., whether the task requires a single step to be performed, or dual, or multiple simultaneous steps, or whether reasoning demands are low or absent, versus high), from the learners’ perceptions of task difficulty, which are a result of the abilities they bring to the task (e.g., intelligence) as well as affective responses (e.g., anxiety). I distinguish both of these from task conditions, which are specified in terms of information flow in classroom participation (e.g., one versus two way tasks), and in terms of the grouping of participants (e.g., same versus different gender). This triadic componential framework enables the complex classroom learning situation to be analysed in a manageable way, allowing interactions among these three broad groups of complexity, difficulty and condition factors to be charted. Of particular interest to me is the interaction of difficulty factors, such as aptitude, with the complexity factors manipulated during task design, and the effects of these interactions on performance and learning. The Cognition Hypothesis also motivates a model for sequencing increases in the complexity of task characteristics described in the Triadic Componential Framework. This SSARC model (Simplify, Stabilize-Automatize-Restructure, Complexify) proposes three stages in which versions of tasks are made more complex for learners (Robinson, 2007, 2009, 2010). Relevant publications are given here, and a longer summary of the Cognition Hypothesis abridged from a 2003 paper is given here.

        Since 1993 I have been examining the generalizability of claims made about implicit and explicit learning in cognitive psychology to the specific domain of adult second language learning. This has involved replicating the methodology used to study this area in cognitive psychology, by such researchers as Arthur Reber, but using natural languages as the learning target (Robinson, 1996a,b, 1997a,b, 2002, 2005, 2007, 2010). I have argued that the results of these studies show, contrary to the claims of Reber and Krashen, that implicit and explicit learning processes are not fundamentally different—they are fundamentally similar. Attention and awareness are implicated in all aspects of second language learning, and individual differences in the cognitive abilities learners bring to the learning condition (implicit, incidental or explicit), or learning task, will directly affect the efficiency of attention allocation, rehearsal of attended information, awareness of it, and so the extent of long term learning. That is, there are aptitudes for implicit, and incidental learning, as well as explicit learning (Robinson, 2001, 2002, 2005, 2007) (as well as aptitudes for learning while performing pedagogic tasks having one versus another characteristic).
        This hypothesis is thus compatible with the claims of Schmidt’s ‘noticing’ hypothesis, which claims awareness is necessary for L2 learning, in contrast to Krashen’s claims about unconscious L2 ‘acquisition’. It is compatible with transfer appropriate processing accounts of explicit and implicit learning, such as those of Roediger (1989), and related episodic-processing accounts such as those of Whittlesea and Dorken (1993). It is also compatible with the work of Perruchet, Shanks, and Dulany and other critics of Reber’s claims that implicit, unconscious learning is necessary to successful learning where a stimulus domain is complex; is insensitive to individual differences in measures of intellectual abilities; and results in an abstract tacit knowledge system that is inaccessible to awareness. In contrast I, like they, argue the results of implicit learning are explainable as the accumulation of consciously processed, fragmentary, associative knowledge. Implicit learning in this view is the accumulation of memory for instances of bigram , trigram and other chunks in long term memory as a result of attended exposure. Some relevant publications are given here, and two summaries of the Fundamental Similarity Hypothesis are given here.

        Arising out of the work on L2 pedagogic tasks and their processing demands, and the effects of experimental L2 learning conditions described above, since around 1998 I have been arguing that Richard Snow’s (1987, 1994) notion of aptitude complexes should be adapted in SLA aptitude research, and that aptitude-treatment interaction research of the kind Snow proposed will be necessary if a complete picture of the effects of specific instructional interventions during classroom learning is to emerge. I argue that aptitude for L2 learning needs to be examined in relation to the processing demands of different learning conditions and tasks, and that different complexes of cognitive abilities are involved in aptitude for learning under these different processing conditions. That is, like Sternberg (1985, 1990, 2002) and Snow (1994), I claim that there are multiple aptitudes for L2 learning, and that previous measures of aptitude, such as Carroll and Sapon’s Modern Language Aptitude Test, are not sufficiently sensitive to capture the interaction of cognitive abilities with the processing demands of contemporary classroom learning conditions and pedagogic interventionist techniques for focus on form.
        While one aim of this research is to further clarify the findings from effects of instruction studies (see e.g., Doughty & Wiliams, 1998; Norris & Ortega, 2000, 2006), by controlling for important learner differences, another applied aim of this research is to match aptitude profiles to the optimum learning conditions, tasks and techniques for inducing focus on form for learners (Long, 1991; Long & Robinson, 1998). I also argue that abilities, or complexes of abilities are much more differentiated in some learners than in others, and that it is especially important to match learners with widely differing strengths and abilities in aptitude complexes to the learning conditions, tasks or techniques most suited to their abilities. Some relevant publications on the Aptitude Complex/ Ability Differentiation framework are given here, and a longer summary is given here.