Over the last decade, fully distributed models have become dominant in connectionist psychological modelling, whereas the virtues of localist models have been underestimated. This target article illustrates some of the benefits of localist modelling. Localist models are characterized by the presence of localist representations rather than the absence of distributed representations. A generalized localist model is proposed that exhibits many of the properties of fully distributed models. It can be applied to a number of problems that are difficult for fully distributed models, and its applicability can be extended through comparisons with a number of classic mathematical models of behaviour. There are reasons why localist models have been underused, though these often misconstrue the localist position. In particular, many conclusions about connectionist representation, based on neuroscientific observation, can be called into question. There are still some problems inherent in the application of fully distributed systems and some inadequacies in proposed solutions to these problems. In the domain of psychological modelling, localist modelling is to be preferred.

Almost all representations have both distributed and localist aspects, depending upon what properties of the data are being considered. With noisy data, features represented in a localist way can be detected very efficiently, and in binary representations they can be counted more efficiently than those represented in a distributed way. Brains operate in noisy environments, so the localist representation of behaviourally important events is advantageous, and fits what has been found experimentally. Distributed representations require more neurons to perform as efficiently, but they do have greater versatility.

Page's target article presents an argument for the use of localist, connectionist models in future psychological theorising. The “manifesto” marshalls a set of arguments in favour of localist connectionism and against distributed connectionism, but in doing so misses a larger argument concerning the level of psychological explanation that is appropriate to a given domain.

Two additional sources of evidence are provided in support of localist coding within connectionist networks. First, only models with localist codes can currently represent multiple pieces of information simultaneously or represent order among a set of items on-line. Second, recent priming data appear problematic for theories that rely on distributed representations. However, a faulty argument advanced by Page is also pointed out.

In his target article Page proposes a definition of the term “localist.” In this commentary I argue that his definition does not serve to make a principled distinction, as the inclusion of vague terms make it susceptible to some problematic counterexamples.

Page proposes a simple, localist, lateral inhibitory network for implementing a selection process that approximately conforms to the Luce choice rule. I describe another localist neural mechanism for selection in accordance with the Luce choice rule. The mechanism implements an independent race model. It consists of parallel, independent nerve fibers connected to a winner-take-all cluster, which records the winner of the race.

Distributed representations can be distributed in very many ways. The specific choice of representation for a specific model is based on considerations unique to the area of study. General statements about the effectiveness of distributed models are therefore of little value. The popularity of these models is discussed, particularly with respect to reporting conventions.

In order to benefit from the advantages of localist coding, neural models that feature winner-take-all representations at the top level of a network hierarchy must still solve the computational problems inherent in distributed representations at the lower levels.

In the Localist Manifesto, Page enumerated several computational advantages that localist representations have over distributed representations, but the most important difference between such networks concerns their theoretical clarity. Distributed representations are normally closed to theoretical interpretation and, for that reason, contribute little to psychology, whereas the meaning of the information processing in networks using localist representations can be transparent.

We contend that if efficiency and reliability are important factors in neural information processing then distributed, not localist, representations are “evolution's best bet.” We note that distributed codes are the most efficient method for representing information, and that this efficiency minimizes metabolic costs, providing adaptive advantage to an organism.

Two categorization arguments pose particular problems for localist connectionist models. The internal representations of localist networks do not reflect the variability within categories in the environment, whereas networks with distributed internal representations do reflect this essential feature of categories. We provide a real biological example of perceptual categorization in the monkey that seems to require population coding (i.e., distributed internal representations).

We focus on two components of Page's argument in favour of localist representations in connectionist networks: First, we take issue with the claim that localist representations can give rise to generalisation and show that whenever generalisation occurs, distributed representations are involved. Second, we counter the alleged shortcomings of distributed representations and show that their properties are preferable to those of localist approaches.

The notion of a “familiar example” used in Page's definition of a “localist model” is shown to be meaningful only with respect to the types of tasks faced by the connectionist model. It is also shown that the modeling task ultimately dictates which choice of model: “localist” or “distributed” is most appropriate.

A number of examples are given of how localist models may incorporate distributed representations, without the types of nonlocal interactions that often render distributed models implausible. The need to analyze the information that is encoded by these representations is also emphasized as a metatheoretical constraint on model plausibility.

An extensive survey by Heathcote et al. (in press) found that the Law of Practice is closer to an exponential than a power form. We show that this result is hard to obtain for models using leaky competitive units when practice affects only the input, but that it can be accommodated when practice affects shunting self-excitation.

Page argues convincingly for several important properties of localist representations in connectionist models of cognition. I argue that another important property of localist representations is that they serve as the starting point for connectionist representations of symbolic (relational) structures because they express meaningful properties independent of one another and their relations.

Page's target article makes a good case for the strength of localist models. This can be characterized as an issue of where new information is integrated with respect to existing knowledge structures. We extend the analysis by discussing the dimension of when this integration takes place, the implications, and how they guide us in the creation of cognitive models.

Page's definition of localism is inspired by the instance-based paradigm. However, the locality of representations is not necessary for a model to be instance-based and, on the other hand, explicit featural representations are generally considered local. The important distinction is between instance-based and noninstance-based paradigms and not between distributed and local representations as Page claims.

Page argues that localist models can be applied to a number of problems that are difficult for distributed models. However, it is easy to find examples where the opposite is true. This commentary illustrates the superiority of distributed models in the domain of artificial grammar learning, a paradigm widely used to investigate implicit learning.

Examining the implications of a localist model for linguistic performance, I show the strengths of the P-graph, a network of elementary units of meaning where utterance results from relaxation through the operation of a dynamics of affect values. A unit of meaning is stored in a synaptic connection that brings together two words. Such a model, consistent with the anatomy and physiology of the neural tissue, eschews a number of traditional pitfalls of “semantic networks”: (1) ambiguity ceases to be an issue, as similar uses of a word are automatically clustered together; (2) faster retrieval of words acquired early is explained by the larger number of their instances. In addition the P-graph takes advantage of a plausible form of information storage: the local topology of the neural tissue.