**3.3 Results – Classification and recognition**

Classification errors were again rare, averaging between 2% on the immediate test following systematic training and visual testing to 11.0% on the delayed test following random training and a haptic test.

On the transfer test, subjects were either provided with the objects to be recognized and classified, based only on its visual appearance, from touch alone, or with both vision and touch provided. As was the case in learning, when an old object was presented to both modalities, the object matched its training pairing. Finally, as was the case in Experiment 1, objects were learned in a systematic or random manner, with testing occurring either

The learning phase again consisted of a series of four 4 study-test trials with corrective feedback. On each learning trial, the participant visually perceived an object of a category (e.g., A1) and at the same time haptically explored, under an opaque black foam board, another object of the same category (e.g., A15). Presentation order for the systematic training condition again presented the objects blocked by category; in the random condition, category pairing was maintained but randomly selected in terms of the category presented. Following a given study block, the objects were randomly presented and the subject was asked to identify the category. In the visual condition, the objects were presented visually but could not be touched; in the haptic condition, each object could be manipulated but not seen. In the visual + haptic condition, the objects could be inspected both visually and haptically. Following each response, corrective feedback was provided. This procedure was repeated 3 additional study/test times. Participants were only informed of a category label and told to form each category by using both the appearance and felt conformations of each presented object. Participants were instructed to haptically explore and visually perceive the

The transfer phase began either immediately or one week after completion of the learning phase. Participants were instructed to classify each object to its appropriate category learned during training (A, B, or C), and recognize whether this object was old or new using vision only, touch only, or both vision and touch. To each randomly presented object, participants gave a double-response after each presentation, recognition (Old or New) followed by classification (A, B, or C). Response time was self-spaced but restricted to 15 sec and

Figure 6 shows the mean accuracy across learning blocks as a function of order of presentation and modality of test following each study trial. The main effect of learning blocks, order, and modality at test, were significant. In general, performance improved across learning blocks, with systematic presentation again facilitating rate of learning. Learning following visual + haptic test produced faster learning than visual alone or haptic alone (p < .05 in each case, Bonferroni test); visual alone also resulted in significantly fewer

Classification errors were again rare, averaging between 2% on the immediate test following systematic training and visual testing to 11.0% on the delayed test following random

immediately or after a delay of one week.

two conflicting stimuli simultaneously.

feedback was not given during transfer test.

**3.3 Results – Classification and recognition** 

**3.2 Results – Learning** 

errors than haptic alone.

training and a haptic test.

**3.1 Method** 

On the recognition, test, the overall hit and false alarm rates were .794 and 533, respectively, which demonstrated that subjects discriminated old from new objects on the transfer test. The best discrimination occurred when recognition was tested visually (P(Hit) = .860, P(FA) = .532), or when both haptic and visual information were available (P(Hit) = .819, P(FA) = .468); when tested by the haptic modality alone, the difference between hits and false alarms remained significant but the level of discrimination was reduced (P(Hit) = .702, P(FA) = .600). A post-hoc Bonferronni test revealed that recognition discrimination was ordered V+H = V > H. Discrimination between old and new objects was also enhanced by systematic presentation during study, P(Hit) = .791 and P(FA) = .502; following random presentation, these values were P(Hit) = .796 and P(FA) = .565.

Fig. 6. Mean learning rate across trial blocks under conditions of cross-modal conflict

#### **3.4 Discussion**

Classification errors were again rare, averaging between 2% on the immediate test following systematic training and visual testing to 11.0% on the delayed test following random training and a haptic test.

Inter-modal conflict neither retarded learning nor degraded recognition. In fact, learning was speeded slightly by intermodal conflict, with learning rates comparing favorably to those obtained in any of the conditions in Experiment 1. Similarly, classification and later recognition was largely unperturbed by this manipulation. The results do show clear dominance by the visual modality, since recognition accuracy for touch alone, following learning with both modalities present, was significant but substantially reduced relative to recognition based on vision alone or when both vision and haptic information was available. This would suggest that, when both visual and haptic information are simultaneously available in the learning of concepts that the resulting concepts are biased by visual information, with haptic information available but playing a reduced role. Finally, as was

Haptic Concepts 15

Second, when texture is the crucial dimension there should be reliable differences in categorization performance across learning trials and transfer between subjects in the partial and complete experience conditions. The integration of the crucial dimension with its related dimensions should become more difficult, if not impossible, if the related dimensions are not simultaneously provided with the crucial dimension, as when texture is the crucial dimension, as opposed to if one of the related dimensions is provided simultaneously with the crucial dimension, as when length is the crucial dimension. As such, for participants with partial experience, those that studied categories with texture as the crucial dimension should have worse categorization performance in learning compared

> Cat A: Cat B:

participants with partial experience and texture as their crucial dimension.

diagnostic categorical information only when texture was the crucial dimension.

These two predictions would result in little difference in categorization accuracy across learning trials between participants with full experience and length as their crucial dimension, participants with partial experience and length as their crucial dimension, and participants with full experience and texture as their crucial dimension, yet all three of those groups of participants would perform very differently across learning trials from

Subjects received 6 learning trials, the results of which are shown in Figure 8. Overall, learning was as predicted – when length was the critical dimension and learning was partial, learning was unaffected, i.e., being deprived of texture (even though texture and length could also be used to discriminate the categories) did not degrade learning, since length could always be combined with width for categorical separation. Similarly, when texture was critical, it was readily learned in the complete condition but learning was severely retarded in the partial condition. That is partial experience inhibited access to

Fig. 7. Categorical structure in the length-critical condition

**4.1 Method, procedure, and results** 

to participants whose crucial dimension was length.

the case in Experiment 1, false recognition of the midpoint and prototype objects was lower than for the new objects closest to the category prototype.
