Categorical versus coordinate spatial relations: Computational analyses and computer simulations.

Abstract

Results of 4 sets of neural network simulations support the distinction between categorical and coordinate spatial relations representations: (a) Networks that were split so that different hidden units contributed to each type of judgment performed better than unsplit networks; the reverse was observed when they made 2 coordinate judgments. (b) Both computations were more difficult when finer discriminations were required; this result mirrored findings with human Ss. (c) Networks with large, overlapping "receptive fields" performed the coordinate task better than did networks with small, less overlapping receptive fields, but vice versa for the categorical task; this suggests a possible basis for observed cerebral lateralization of the 2 kinds of processing. (d) The previously observed effect of stimulus contrast on this hemispheric asymmetry could reflect contributions of more neuronal input in high-contrast conditions.

Picture the difference: electrophysiological investigations of picture processing in the two cerebral hemispheres

“…absolute distance) (e.g. [29,35,36]). One crucial difference between these types of spatial relations is that the former is reference frame dependent (e.g.…”

Distance and distraction effects in the apprehension of spatial relations.

“…Different vertical positions are associated more or less strongly with above and below, and if position coding were sufficiently coarse, adjacent positions may activate both above and below. Thus, apprehension may become easier as distance increases, as in the Kosslyn et al (1992) model. However, this prediction depends on aspects of the model that Hummel and Biederman (1992) did not specify, so it is less critical to the viability of their model than the prediction about distractors, which they did specify.…”

“…The representations underlying apprehension in the Kosslyn et al (1992) theory are like the spatial template representation in the Logan and Sadler (in press) theory in that they divide space into regions that contain good and bad examples of the relation. Kosslyn et al (1992) go beyond Logan and Sadler in building a three-layer connectionist model of the apprehension process, which they trained to discriminate above from below.…”

“…Distractors may increase the effect of distance because Step c will have to be iterated (i.e., comparing the indexed object with a description of the located object), but the distance effect should be significant even when no distractors are present. Kosslyn, Chabris, Marsolek, and Koenig (1992) Kosslyn, Chabris, Marsolek, and Koenig (1992; see also Kosslyn, 1987;O'Reilly, Kosslyn, Marsolek, & Chabris, 1990;Rueckl, Cave, & Kosslyn, 1989) proposed a connectionist model of apprehension of categorical spatial relations, like above and below, which they contrasted with apprehension of metric or coordinate spatial relations, like those used to navigate between specific obstacles, and with apprehension of object identities. The main force of their work was in contrasting different kinds of apprehension to motivate hypotheses about differential localization of the underlying processes in the brain, but they made predictions about the effects of distance nevertheless.…”

“…Kosslyn et al (1992) go beyond Logan and Sadler in building a three-layer connectionist model of the apprehension process, which they trained to discriminate above from below. The key feature of their model that is relevant to the present article is their prediction that regions close to the reference object are harder to classify as above or below than regions more remote from it (see Kosslyn et al, 1992, Study 2).…”