Highly complex and abstract skills – such as analogous thinking – have frequently been deemed to represent the very pinnacle of human cognition (e.g., Penn et al. Reference Penn, Holyoak and Povinelli2008). Leech et al. appear to accept this view, but they do not appear to be content to revel in it. Instead, they seek to understand the emergence of analogous thinking, thereby making its study a decidedly developmental matter.

Leech et al. hypothesize that analogical completion arises from simple cognitive mechanisms. Specifically, they suggest that relational priming is a basic building block for completing analogies. To the extent that their innovative account is successful, at least some key aspects of analogical reasoning may not require the hypothesization of analogy-specific mechanisms.

Leech et al. further suggest that analogical processing may best be viewed as an umbrella term that comprises different task-specific concatenations of basic memory and control processes. Analogy-specific mechanisms may very well exist, but other possibilities should be entertained first because of their greater parsimony and plausibility for infants and young children. Finally, any shifts in children's strategies of task mastery are believed to be the result of children acquiring greater and richer relational knowledge. Leech et al. thus stress the interaction between learning mechanisms and environmental experiences in determining children's developmental trajectory.

Leech et al.'s approach suggests how an advanced cognitive competence – such as analogy formation and performance – can be grounded in more elementary processes, and it promises to provide a fuller picture of the mechanisms underlying the transition from simple to more complex reasoning. If there is at least a seed of truth to Ernst Haeckel's recapitulation theory that “ontogeny recapitulates phylogeny,” then applying Leech et al.'s approach to the evolution of analogical behavior might prove to be valuable as well.

For instance, different species of animals appear to be more or less successful in solving a wide range of relational discrimination problems (Wasserman & Zentall Reference Wasserman and Zentall2006). One of the most intensely studied of such relational discrimination problems is same-different discrimination learning (Cook & Wasserman Reference Cook, Wasserman, Wasserman and Zentall2006; Delius Reference Delius, Bertelson and Eelen1994). Here, the behavioral evidence suggests that pigeons, baboons, chimpanzees, and humans all can discriminate first-order same-different relations; they can reliably report whether two or more stimuli are identical (A=A or B=B) or nonidentical (A≠B).

An even more advanced form of same-different discrimination involves higher-order relations between first-order relations. Task mastery here requires organisms to discriminate groups of two or more stimuli that involve the same higher-order relations ([A=A]=[B=B] or [A≠B]=[C≠D]; both groups of stimuli are the same or both groups of stimuli are different) from groups of two or more stimuli that entail different higher-order relations ([A=A]≠[C≠D]; one group of stimuli is the same and the other group of stimuli is different). Such higher-order relations may share important similarities with human analogical reasoning (Thompson & Oden Reference Thompson and Oden2000).

Can only human beings discriminate such higher-order relations and exhibit analogical reasoning? Perhaps not. Premack (Reference Premack1983) and Thompson and Oden (Reference Thompson and Oden2000) have suggested that both humans and apes can appreciate higher-order stimulus relations. Comparative study thus becomes critical in deciding among these and other rival hypotheses and in elucidating the evolutionary origins of analogical thinking.

Such comparative study is already under way. Cook and Wasserman (Reference Cook and Wasserman2007) and Fagot et al. (Reference Fagot, Wasserman and Young2001) have reported that pigeons and baboons, respectively, can discriminate second-order same-different relations in a matching-to-sample task; that discrimination also transfers to novel stimuli. So, the hypothesis that only humans and apes can learn about higher-order same-different relations seems to be too restricted (Premack Reference Premack1983; Thompson & Oden Reference Thompson and Oden2000).

Note that the pigeons and baboons in these two studies were not experimentally naive; each animal had some form of first-order same-different discrimination training prior to second-order relational matching-to-sample training. Perhaps previously learning to attend to simpler first-order relations is a prerequisite to pigeons' and baboons' subsequent success in discriminating more difficult higher-order relations.

Also note that, in both first-order and second-order relational discriminations, pigeons and baboons appeared to rely heavily on the variability of the items in a sample array (Wasserman et al. Reference Wasserman, Young and Cook2004). Young and Wasserman (Reference Young and Wasserman1997) hypothesized that variability or entropy discrimination lies at the root of same-different discrimination (for a mechanistic theory of entropy discrimination, see Young et al. Reference Young, Wasserman and Ellefson2007). Two or more identical items always involve zero entropy; two or more nonidentical items always involve nonzero entropy, with entropy increasing as a direct function of the number of nonidentical items. Not only do pigeons (Young & Wasserman Reference Young and Wasserman1997; Young et al. Reference Young, Wasserman and Garner1997) and baboons (Wasserman et al. Reference Wasserman, Fagot and Young2001a; Reference Wasserman, Young and Fagot2001b) show strong sensitivity to the variability of the items in an array, but so also do college students – in both choice accuracy and choice reaction time (Castro et al. Reference Castro, Young and Wasserman2006; Young & Wasserman Reference Young and Wasserman2001; Reference Young and Wasserman2002).

Hence, we see that the interpretive framework proposed by Leech et al. may be effectively applied to both the development and evolution of relational discrimination behavior. Assuming that complex cognitive processes depend on and progressively evolve from more basic perceptual and cognitive processes – the notion of anagenesis (Gottlieb Reference Gottlieb1984) – we can make the following predictions. First, some species might only be sensitive to the variability of arrays of stimuli. Second, other, more select species may also be able to categorize stimulus arrays into those comprising identical or nonidentical items. Third, other, even more select species may be able to discriminate identical from nonidentical relations. Finally, we might expect to see this task ordering preserved in the development of relational responding in individual animals and to find this ordering to strongly depend on the animal's environmental experiences.

There's no “flying into flying” according to this proposal for the development and evolution of relational responding!