Horne and Lowe (1996) define naming as an intra-individual, circular category relation among stimuli (even those that have no physical features in common) that evoke the same speaker-listener behaviour, and within which, novel behaviour, trained to one stimulus, may generalise, without direct training, to the others. Name-based categorising and novel
behaviour transfer were investigated in 2.5 to 4 year old children. In Experiment l(a), the participants were trained to say "Vek" to each of three, and "Zag" to each of the other three, among six differently-shaped stimuli that were then presented in a Category Matchto-Sample test; the experimenter selected one stimulus in tum to serve as the sample, and
asked each participant to give "the others" from the remaining five. For children who failed, a repeat test, in which they were required to name the sample before selecting the others, was conducted. In Experiment l(b), a subgroup of the children repeated the procedure with a new set of stimuli; categorising of both sets combined (12 stimuli) was next tested, and again at a six-week follow-up. With new participants, in Experiments 2(a) and 2(b), the procedure was as for Experiments l(a) and l(b), respectively, except that the trained common speaker behaviours were manual responses. In Experiment 3(a),
transfer of manual behaviours, trained to a subset of the stimuli in the Set 1 common vocal name relation, was tested; a common manual name relation was trained for Set 2, then tested for transfer of the Set 1 vocal responses. In Experiment 3(b ), the modalities of training and testing were the converse of those in Experiment 3(a). In Experiment 3(c), six-member speaker relations, trained in one modality (either vocal or manual), were tested for transfer of novel behaviour in the alternative modality. The data from these experiments were consistent with the naming account.