Holger Schultheis

Spatial Cognition: Reasoning, Action, Interaction

The Transregional Collaborative Research Center SFB/TR 8 Spatial Cognition pursues interdisciplinary research on a broad range of topics related to the representation and processing mechanisms for intelligent spatial behavior in technical and in natural systems. This contribution gives an overview of the field of research worked on in the SFB/TR 8 Spatial Cognition and presents three representative examples that illustrate the activities in the three research areas Reasoning, Action, and Interaction.

Casimir: An Architecture for Mental Spatial Knowledge Processing

Mental spatial knowledge processing often uses spatio-analogical or quasipictorial representation structures such as spatial mental models or mental images. The cognitive architecture Casimir is designed to provide a framework for computationally modeling human spatial knowledge processing relying on these kinds of representation formats. In this article, we present an overview of Casimir and its components. We briefly describe the long-term memory component and the interaction with external diagrammatic representations. Particular emphasis is placed on Casimirs working memory and control mechanisms. Regarding working memory, we describe the conceptual foundations and the processing mechanisms employed in mental spatial reasoning. With respect to control, we explain how it is realized as a distributed, emergent facility within Casimir.

ALTSim: A MATLAB simulator for current associative learning theories

ALTSim is a MATLAB-based simulator of several associative learning models, including Pearce’s configural model, the extended configural model, the Rescorla-Wagner model, the unique cue hypothesis, the modified unique cue hypothesis, the replaced elements model, and Harris’s elemental model. It allows for specifying all relevant parameters, as well as exact stimulus sequences by graphical user interfaces. It is an easy-to-use tool that facilitates evaluating and comparing the featured associative learning models. ALTSim is available free of charge from www.staff.uni-marburg.de/~lachnit/ALTSim/.We study the problem of learning a non-parametric mapping between two continuous spaces without having access to input-output pairs for training, but rather to groups of input-output pairs, where the correspondence structure within each group is unknown and where outliers may be present. This problem is solved by transforming each space using the channel representation, and finding a linear mapping on the transformed domain. The asymptotical behavior of the method for a large number of training samples is found to be very related to the case of known correspondences. The results are evaluated on simulated data

Rapid-REM: a MATLAB simulator of the replaced-elements model.

A recent proposal for an elemental account of associative learning phenomena is the replaced-elements model (REM) put forward by Wagner (2003). Although the ideas underlying this model are comparatively simple, implementation of the model is rather complex. In this article, we present Rapid-REM, a MATLAB simulator of Wagner’s model. Rapid-REM features a graphical user interface for manipulating all essential parameter values and for control of the simulation process, graphical visualization of the simulation course and the results, and the alternative possibility of simulating the replaced-elements model as it was originally proposed (Wagner & Brandon, 2001). Rapid-REM is available free of charge from www.staff.uni-marburg.de/≈lachnit/Rapid-REM/. This simulator makes it easy to derive predictions for REM and evaluate them, and it will therefore facilitate insights into the mechanisms of associative learning.

The spatial and the visual in mental spatial reasoning: an ill-posed distinction

It is an ongoing and controversial debate in cognitive science which aspects of knowledge humans process visually and which ones they process spatially. Similarly, artificial intelligence (AI) and cognitive science research, in building computational cognitive systems, tended to use strictly spatial or strictly visual representations. The resulting systems, however, were suboptimal both with respect to computational efficiency and cognitive plausibility. In this paper, we propose that the problems in both research strands stem from a misconception of the visual and the spatial in mental spatial knowledge processing. Instead of viewing the visual and the spatial as two clearly separable categories, they should be conceptualized as the extremes of a continuous dimension of representation. Regarding psychology, a continuous dimension avoids the need to exclusively assign processes and representations to either one of the categories and, thus, facilitates a more unambiguous rating of processes and representations. Regarding AI and cognitive science, the concept of a continuous spatial / visual dimension provides the possibility of representation structures which can vary continuously along the spatial / visual dimension. As a first step in exploiting these potential advantages of the proposed conception we (a) introduce criteria allowing for a non-dichotomic judgment of processes and representations and (b) present an approach towards representation structures that can flexibly vary along the spatial / visual dimension.

Modeling Mental Spatial Reasoning About Cardinal Directions

This article presents research into human mental spatial reasoning with orientation knowledge. In particular, we look at reasoning problems about cardinal directions that possess multiple valid solutions (i.e., are spatially underdetermined), at human preferences for some of these solutions, and at representational and procedural factors that lead to such preferences. The article presents, first, a discussion of existing, related conceptual and computational approaches; second, results of empirical research into the solution preferences that human reasoners actually have; and, third, a novel computational model that relies on a parsimonious and flexible spatio-analogical knowledge representation structure to robustly reproduce the behavior observed with human reasoners.

HMS: a MATLAB simulator of the Harris model of associative learning.

Harris (2006) recently proposed a new elemental model of the processes involved in associative learning. Although Harris explicated all relevant mathematical and conceptual details of the model in his article, implementing a computer simulation of his model requires considerable programming expertise and work. We therefore present the Harris model simulator (HMS), a MATLAB simulator of Harris’s model. HMS provides a graphical user interface for manipulating all essential parameter values and for controlling the simulation process, the graphical visualization of the simulation course, and the numerical results. HMS is available free of charge from www.staff.uni-marburg.de/≈lachnit/harris/. HMS allows researchers to easily derive and evaluate predictions for the Harris model, and it will therefore facilitate insights into the mechanisms of associative learning.

Differences between Spatial and Visual Mental Representations

This article investigates the relationship between visual mental representations and spatial mental representations in human visuo-spatial processing. By comparing two common theories of visuo-spatial processing – mental model theory and the theory of mental imagery – we identified two open questions: (1) which representations are modality-specific, and (2) what is the role of the two representations in reasoning. Two experiments examining eye movements and preferences for under-specified problems were conducted to investigate these questions. We found that significant spontaneous eye movements along the processed spatial relations occurred only when a visual mental representation is employed, but not with a spatial mental representation. Furthermore, the preferences for the answers of the under-specified problems differed between the two mental representations. The results challenge assumptions made by mental model theory and the theory of mental imagery.

Determinants of attentional modulation near the hands

A series of visual search experiments conducted by Abrams et al. (2008) indicates that disengagement of visual attention is slowed when the array of objects that are to be searched are close to the hands (hands on the monitor) than if they are not close to the hands (hands in the lap). These experiments establish the impact ones hands can have on visual attentional processing. In the current paper we more closely examine these two hand postures with the goal of pinpointing which characteristics are crucial for the observed differences in attentional processing. Specifically, in a set of 4 experiments we investigated additional hand postures and additional modes of response to address this goal. We replicated the original Abrams et al. (2008) effect when only the two original postures were used; however, surprisingly, the effect was extinguished with the new range of postures and response modes, and this extinction persisted across different populations (German and English students), and different experimental hardware. Furthermore, analyses indicated that it is unlikely that the extinction of the effect was caused by increased practice due to additional blocks of trials or by an increased probability that participants were able to guess the purpose of the experiment. As such our results suggest that in addition to the nature of the postures of the hand, the number of postures is a further important factor that influences the impact the hands have on visual processing.

Attentional Distribution and Spatial Language

Whether visual spatial attention can be split to several discontinuous locations concurrently is still an open and intensely debated question. We address this question in the domain of spatial language use by comparing two existing and three newly proposed computational models. All models are assessed regarding their ability to account for human acceptability ratings for how well a given spatial term describes the spatial arrangement of two functionally related objects. One of the existing models assumes that taking the functional relations into account involves split attention. All new models incorporate functional relations without assuming split attention. Our simulations suggest that not assuming split attention is more appropriate for taking the functional relations into account than assuming split attention. At the same time, the simulations raise doubt as to whether any of the models appropriately captures the impact of functional relations on spatial language use.