Where the isocortex comes from is an important question, but even more important is understanding what it leads to – that is, what advantage is afforded by its peculiar organization in layers of distinct neuronal types. A computational hypothesis accounts for granulation and for the differentiation between supra- and infragranular pyramidal layers, as quantitatively advantageous to support fine topography in sensory maps.

A dorsalization mechanism is a good candidate for the evolutionary origin of the isocortex, producing a radial and tangential expansion of the dorsal pallium (and perhaps other structures that acquired a cortical phenotype). Evidence suggests that a large part of the dorsal ventricular ridge (DVR) of reptiles and birds derives from the embryonic ventral pallium, whereas the isocortex possibly derives mostly from the dorsal pallium. In early mammals, the development of olfactory-hippocampal associative networks may have been pivotal in facilitating the selection of a larger and more complex dorsal pallium which received both collothalamic and lemnothalamic sensory information. Finally, although it is not clear exactly when mammalian brain expansion began, fossil evidence indicates that this was a late event in mammaliaform evolution.

Newell (1980; 1990) proposed that cognitive theories be developed in an effort to satisfy multiple criteria and to avoid theoretical myopia. He provided two overlapping lists of 13 criteria that the human cognitive architecture would have to satisfy in order to be functional. We have distilled these into 12 criteria: flexible behavior, real-time performance, adaptive behavior, vast knowledge base, dynamic behavior, knowledge integration, natural language, learning, development, evolution, and brain realization. There would be greater theoretical progress if we evaluated theories by a broad set of criteria such as these and attended to the weaknesses such evaluations revealed. To illustrate how theories can be evaluated we apply these criteria to both classical connectionism (McClelland & Rumelhart 1986; Rumelhart & McClelland 1986b) and the ACT-R theory (Anderson & Lebiere 1998). The strengths of classical connectionism on this test derive from its intense effort in addressing empirical phenomena in such domains as language and cognitive development. Its weaknesses derive from its failure to acknowledge a symbolic level to thought. In contrast, ACT-R includes both symbolic and subsymbolic components. The strengths of the ACT-R theory derive from its tight integration of the symbolic component with the subsymbolic component. Its weaknesses largely derive from its failure, as yet, to adequately engage in intensive analyses of issues related to certain criteria on Newell's list.

Newell wanted a theory of cognition to abide by some explicit criteria, here called the Newell Test. The test differs from the Turing Test because it is explicit. The Newell Test will include the Turing Test if its characterization of cognition is complete. It is not. Its use here is open-ended: A system that does not pass it well invites improvement.

The notion of functionality is appropriately central to the Newell Test but is also critical at a lower level, in development of cognitive sub-theories. I illustrate, on one hand, how far this principle is from general acceptance among verbal theoreticians, and, on the other hand, how simulation models (here implemented within ACT-R) seem to drive the functional question automatically.

In this commentary I explore nonclassical connectionism (NCC) as a coherent framework for evaluation in the spirit of the Newell Test. Focusing on knowledge integration, development, real-time performance, and flexible behavior, I argue that NCC's “within-theory rank ordering” would place subsymbolic modeling in a better position. Failure to adopt a symbolic level of thought cannot be interpreted as a weakness.

“Conceptual coordination” analysis bridges connectionism and symbolic approaches by positing a “process memory” by which categories are physically coordinated (as neural networks) in time. Focusing on dysfunctions and odd behaviors, like slips, reveals the function of consciousness, especially constructive processes that are often taken for granted, which are different from conventional programming constructs. Newell strongly endorsed identifying architectural limits; the heuristic of “diagnose unusual behaviors” will provide targets of opportunity that greatly strengthen the Newell Test.

We distinguish traditional cognition theories from hierarchically complex stacked neural networks that meet many of Newell's criteria. The latter are flexible and can learn anything that a person can learn, by using their mistakes and successes the same way humans do. Shortcomings are due largely to limitations of current technology.

I have three types of interrelated comments. First, on the choice of the proposed criteria, I argue against any list and for a system of criteria. Second, on grading, I suggest modifications with respect to consciousness and development. Finally, on the choice of “theories” for evaluation, I argue for Edelman's theory of neuronal group selection instead of connectionism (classical or not).

We use the Newell Test as a basis for evaluating ACT-R as an effective architecture for cognitive engineering. Of the 12 functional criteria discussed by Anderson & Lebiere (A&L), we discuss the strengths and weaknesses of ACT-R on the six that we postulate are the most relevant to cognitive engineering.

ACT is compared with a particular type of connectionist model that cannot handle symbols and use nonbiological operations which do not learn in real time. This focus continues an unfortunate trend of straw man debates in cognitive science. Adaptive Resonance Theory, or ART-neural models of cognition can handle both symbols and subsymbolic representations, and meet the Newell criteria at least as well as connectionist models.

We share with Anderson & Lebiere (A&L) (and with Newell before them) the goal of developing a domain-general framework for modeling cognition, and we take seriously the issue of evaluation criteria. We advocate a more focused approach than the one reflected in Newell's criteria, based on analysis of failures as well as successes of models brought into close contact with experimental data. A&L attribute the shortcomings of our parallel-distributed processing framework to a failure to acknowledge a symbolic level of thought. Our framework does acknowledge a symbolic level, contrary to their claim. What we deny is that the symbolic level is the level at which the principles of cognitive processing should be formulated. Models cast at a symbolic level are sometimes useful as high-level approximations of the underlying mechanisms of thought. The adequacy of this approximation will continue to increase as symbolic modelers continue to incorporate principles of parallel distributed processing.

This commentary questions the applicability of the Newell Test for evaluating the utility of connectionism. Rather than being a specific theory of cognition (because connectionism can be used to model nativist, behaviorist, or constructivist theories), connectionism, we argue, offers researchers a collection of computational and conceptual tools that are particularly useful for investigating and rendering specific fundamental issues of human development. These benefits of connectionism are not well captured by evaluating it against Newell's criteria for a unified theory of cognition.

One supposition underlying the Anderson & Lebiere (A&L) target article is that the maximally broad “encompassing of its subject matter – the behavior of man” (cf. sect. 6, last para.) is regarded as an unquestioned quality criterion for guiding cognitive research. One might argue for an explicit specification of the limitations of a given paradigm, rather than extending it to apply to as many domains as possible.

The target article is based upon the principle that complex mental phenomena result from the interactions among some elementary entities. Connectionist nodes and ACT-R's production rules can be considered as such entities. However, before testing against Newell's macro-criteria, self-organizing models must be tested against criteria relating to the properties of their elementary entities. When such micro-criteria are considered, they separate connectionism from ACT-R and the comparison of these theories against Newell's Tests is hardly correct.

The distinction between such differing approaches to cognition as connectionism and rule-based models is paralleled by a distinction between two basic modes of cognition postulated in the so-called dual-process theories. Integrating these theories with insights from hybrid systems might help solve the dilemma of combining the demands of evolutionary plausibility and computational universality. No single approach alone can achieve this.

This commentary defines an additional characteristic of human learning. The nature of this test is different from the ones by Newell: This is a hard, pass/fail type of test. Thus a theory of cognition cannot partially satisfy this test; it either conforms to the requirement fully, or it doesn't. If a theory of cognition cannot satisfy this property of human learning, then the theory is not valid at all.

The Newell Test is an ambitious and promising project, but not without pitfalls. Some of the current criteria are not theoretically neutral, whereas others are unhelpful. To improve the test, the learning and development criteria are reviewed and revised, which suggests adding a maturation criterion as well. Such changes should make the Newell Test more general and useful.

I present reasons for adding an embodiment criterion to the list defended by Anderson & Lebiere (A&L). I also entertain a likely objection contending that embodiment is merely a type of dynamic behavior and is therefore covered by the target article. In either case, it turns out that neither connectionism nor ACT-R do particularly well when it comes to embodiment.

This commentary examines one aspect of the target article – the comparison of ACT-R with connectionist models. It argues that conceptions of connectionist models should be broadened to cover the whole spectrum of work in this area, especially the so-called hybrid models. Doing so may change drastically ratings of connectionist models, and consequently shed more light on the developing field of cognitive architectures.