The development of analogical reasoning has traditionally been understood in terms of theories of adult competence. This approach emphasizes structured representations and structure mapping. In contrast, we argue that by taking a developmental perspective, analogical reasoning can be viewed as the product of a substantially different cognitive ability – relational priming. To illustrate this, we present a computational (here connectionist) account where analogy arises gradually as a by-product of pattern completion in a recurrent network. Initial exposure to a situation primes a relation that can then be applied to a novel situation to make an analogy. Relations are represented as transformations between states. The network exhibits behaviors consistent with a broad range of key phenomena from the developmental literature, lending support to the appropriateness of this approach (using low-level cognitive mechanisms) for investigating a domain that has normally been the preserve of high-level models. Furthermore, we present an additional simulation that integrates the relational priming mechanism with deliberative controlled use of inhibition to demonstrate how the framework can be extended to complex analogical reasoning, such as the data from explicit mapping studies in the literature on adults. This account highlights how taking a developmental perspective constrains the theory construction and cognitive modeling processes in a way that differs substantially from that based purely on adult studies, and illustrates how a putative complex cognitive skill can emerge out of a simple mechanism.

According to the target article authors, initial experience with a circumstance primes a relation that can subsequently be applied to a different circumstance to draw an analogy. While I broadly agree with their claim about the role of relational priming in early analogical reasoning, I put forward a few concerns that may be worthy of further reflection.

The target article criticises neural-symbolic systems as inadequate for analogical reasoning and proposes a model of analogy as transformation (i.e., learning). We accept the importance of learning, but we argue that, instead of conflicting, integrated reasoning and learning would model analogy much more adequately. In this new perspective, modern neural-symbolic systems become the natural candidates for modelling analogy.

This commentary questions the general assumptions concerning the cognitive value of analogical reasoning on which the argument developed by Leech et al. appears to rest. In order to better assess the findings of their meta-analysis, it shifts the perspective from development to evolution, and frames their “how” concern within a broader “why” issue.

Despite its strengths, Leech et al.'s model fails to address the important benefits that derive from self-explanation and task feedback in analogical reasoning development. These components encourage explicit, self-reflective processes that do not necessarily link to knowledge accretion. We wonder, therefore, what mechanisms can be included within a connectionist framework to model self-reflective involvement and its beneficial consequences.

Leech et al. make two proposals: that relational priming is central to analogy, and that relations between objects are best represented as transformations of those objects. Although their account of analogy as relational priming is a useful contribution to our understanding of analogical development, in this commentary I show that relations in general cannot be represented by transformations.

Six questions are posed that are really specific versions of this question: How can Leech et al.'s system be extended to handle adult-level analogies that frequently combine concepts from semantically distant domains sharing few relational labels and that involve the production of abstractions? It is Leech et al. who stress development; finding such an extension would seem to have to be high on their priority list.

Leech et al.'s model proposes representing relations as primed transformations rather than as structured representations (explicit representations of relations and their roles dynamically bound to fillers). However, this renders the model unable to explain several developmental trends (including relational integration and all changes not attributable to growth in relational knowledge). We suggest looking to an alternative computational model that learns structured representations from examples.

We evaluate whether evidence from conceptual combination supports the relational priming model of analogy. Representing relations implicitly as patterns of activation distributed across the semantic network provides a natural and parsimonious explanation of several key phenomena observed in conceptual combination. Although an additional mechanism for role resolution may be required, relational priming offers a promising approach to analogy.

Relational priming is argued to be a deeply inadequate model of analogy-making because of its intrinsic inability to do analogies where the base and target domains share no common attributes and the mapped relations are different. Leech et al. rely on carefully handcrafted representations to allow their model to make a complex analogy, seemingly unaware of the debate on this issue fifteen years ago. Finally, they incorrectly assume the existence of fixed, context-independent relations between objects.

Leech et al.'s demonstration that analogical reasoning can be an emergent property of low-level incremental learning processes is critical for analogical theory. Along with insights into neural learning based on the salience of dynamic spatio-temporal structure, and the neural priming mechanism of repetition suppression, it establishes relational primacy as a plausible theoretical description of how brains make analogies.

The authors propose that analogical reasoning may be achieved without conscious or explicit deliberation. The argument would be strengthened by more convincingly demonstrating instances of analogy that do not require explicit deliberation. Recent findings demonstrate that deliberative or explicit strategies are not necessary for flexible expression under novel circumstances (Greene et al. 2001) to include analogical transfer (Gross & Greene 2007). This issue is particularly critical because the existence of relational priming poses a serious challenge to the widely held notion that flexible expression of learned relations requires deliberative processes.

Humans, including preschool children, exhibit role-based relational reasoning, of which analogical reasoning is a canonical example. The “role-less” connectionist model proposed in the target article is only capable of conditional paired-associate learning.

Analogy-making is an ability with which we can abstract from surface similarities and perceive deep, meaningful similarities between different mental objects and situations. I propose that mental objects are dynamically changing sets of potentially interchangeable connotations. Unfortunately, most models of analogy seem devoid of both semantics and relevance-extraction, postulating analogy as a one-to-one mapping devoid of connotation transfer.

It is important to take a developmental approach to the problem of analogy. One limitation of this approach, however, is that it does not deal with the complexity of making analogical inferences. There are a few key principles of analogical inference that are not well captured by the analogical relational priming (ARP) model.

Analogy employs a neurocognitive working-memory (WM) system to activate and bind relational representations, integrate multiple relations, and suppress distracting information. Analogy experiments exploring these processes have used a variety of methodologies including dual tasks, neuropsychology, and functional neuroimaging, as well as experiments with children and older adults. Collectively, these experiments provide a rich set of results useful in evaluating any model of analogy and its development.

Leech et al.'s analysis adds to an emerging consensus of the role of priming in analogy-making. However, their model cannot scale up to adult-level performance because not all relations can be cast as functions. One-size-fits-all accounts cannot capture the richness of analogy. Proportional analogies and transitive inferences can be made by nonstructural mechanisms. Therefore, these tasks do not generalize to tasks that require structure mapping.

Analogy by priming learned transformations of (causally) related objects fails to explain an important class of inference involving abstract source-target relations. This class of analogical inference extends to ad hoc relationships, precluding the possibility of having learned them as object transformations. Rather, objects may be placed into momentarily corresponding, symbolic, source-target relationships just to complete an analogy.

The authors' claim that analogical reasoning is the product of relational priming is compatible with language processing work that emphasizes the role of low-level automatic processes in the alignment of situation models in dialogue. However, their model ignores recent behavioral evidence demonstrating a “lexical boost” effect on relational priming. We discuss implications of these data.

The target article develops a computational connectionist model for analogy-making from a developmental perspective and evaluates this model using simple analogies. Our commentary critically reviews the advantages and limits of this approach, in particular with respect to its expressive power, its capability to generalize across analogous structure and analyze systematicity in analogies.