Behavioural Brain Research 149 (2004) 107–111 Short communication Development of configural 3D object recognition Ingo Rentschler a,∗ , Martin Jüttner b , Erol Osman a , Alexander Müller a , Terry Caelli c a Institute of Medical Psychology, University of Munich, Goethestrasse 31, 80336 Munich, Germany b Neuroscience Research Institute, School of Life & Health Sciences—Psychology, Aston University, Aston Triangle, Birmingham B4 7ET, UK c Department of Computing Science, University of Alberta, Edmonton, Alta., Canada T6G 2H1 Received 9 May 2003; received in revised form 29 May 2003; accepted 3 June 2003 Abstract There is evidence for the late development in humans of configural face and animal recognition. We show that the recognition of artificial three-dimensional (3D) objects from part configurations develops similarly late. We also demonstrate that the cross-modal integration of object information reinforces the development of configural recognition more than the intra-modal integration does. Multimodal object representations in the brain may therefore play a role in configural object recognition. © 2003 Elsevier B.V. All rights reserved. Keywords: Object recognition; Representation; Configural recognition; Mirror symmetry; Category learning; Cross-modal; Multimodal; Cognitive development 1. Introduction For retrieving the spatial structure of three-dimensional (3D) objects from single static two-dimensional (2D) views, humans require additional information. It has been suggested that this is accomplished by relating image features to visual representations in the brain [2,4]. Yet object knowledge re- tains an intrinsic multimodal quality as it may be acquired both by vision and by touch. The haptic modality can be as- sumed to yield object information as rich as that provided by the visual modality [15], and haptics and vision tend to complement one another in the type of information used to represent the object [17,18,24]. This indicates that 3D object recognition may benefit from the multimodal integration of sensory information. Evidence for the existence of multimodal object rep- resentations in the brain comes from neurophysiological research including functional magnetic resonance imag- ing (fMRI) and transcranial magnetic stimulation (TMS) techniques. TMS experiments suggest that the visual cor- tex is also closely involved in the tactile discrimination of object orientation and shape [33]. fMRI studies probing cross-modal priming in object recognition have revealed an activation not only in the somatosensory cortex but also in ∗ Corresponding author. Tel.: +49-89-5996-641; fax: +49-89-7499-6873. E-mail address: ingo@imp.med.uni-muenchen.de (I. Rentschler). regions of the occipital cortex traditionally associated with visual processing, thus pointing at overlapping haptic and visual representations [1,14]. From developmental psychology, it would seem that mul- timodal object recognition occurs already in infants [20]. The finding, however that infants can use a variety of cues to perceive objects [22,31] does not answer the questions as to when and “how these cues (i.e. knowledge sources) are combined in the process of constructing coherent ob- ject representations” ([27], p. 32). Experiments on visual animal recognition [10] indicate that recognition abilities in young children (under the age of 10 years) draw much upon properties of isolated object parts, whereas compara- ble performance levels with regard to the processing of part configurations, or global form, are reached only in adoles- cence (around 15–16 years). A similar development from the processing of parts to that of part configurations has been postulated for face recognition [5,6,8]. Taken together these observations suggest that the config- ural recognition of artificial 3D objects may occur as late as face and animal recognition do, and that multimodal object representations play a role in its development. To test this hypothesis we compared the developmental characteristics of the effects of haptic and visual prior knowledge on vi- sually learning 3D objects differing in part configurations only. For achieving this goal, we employed a psychophys- ical paradigm based on two essentials. First, a learning set of unfamiliar complex objects containing mirror-symmetric 0166-4328/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0166-4328(03)00194-3