T. B. Ward (1994) investigated creativity by asking participants to draw alien creatures that they imagined to be from a planet very different from Earth. He found that participant drawings reliably contained features typical of common Earth animals. As a consequence, Ward concluded that creativity is structured. The present investigation predicts that this limitation on creativity is not restricted to drawings: the use of different technology will not change creative output. To investigate this question, participants performed Ward's task twice: once using pencil and paper and once using software made to design creatures (the Spore Creature Creator). Only minor significant differences were found. This preliminarily suggests that changing tools does not affect the overall rigidity of the creative process. This lends further support to Ward's thesis on the structural rigidity of creativity. We conclude by suggesting an elaboration to Ward's thesis that will be explored in future work. We suggest that aesthetics might be one of the factors that contribute to creative constraint, in that creatures that are too unusual would be less interesting.

We propose a new algorithm and formal description of generative cognition in terms of the multi-label bagof-words paradigm. The algorithm, Coherence Net, takes its inspiration from evolutionary strategies, genetic programming, and neural networks. We approach generative cognition in spatial reasoning as the decompression of images that were compressed into lossy feature sets, namely, conditional probabilities of labels. We show that the globally parallel and locally serial optimization technique described by Coherence Net is better at accurately generating contextually coherent subsections of the original compressed images than a competitive, purely serial model from the literature: Coherencer.

We propose a training and evaluation approach for autoencoder Generative Adversarial Networks (GANs), specifically the Boundary Equilibrium Generative Adversarial Network (BEGAN), based on methods from the image quality assessment literature. Our approach explores a multidimensional evaluation criterion that utilizes three distance functions: an l1 score, the Gradient Magnitude Similarity Mean (GMSM) score, and a chrominance score. We show that each of the different distance functions captures a slightly different set of properties in image space and, consequently, requires its own evaluation criterion to properly assess whether the relevant property has been adequately learned. We show that models using the new distance functions are able to produce better images than the original BEGAN model in predicted ways.

An incoherent visualization is when aspects of different senses of a word (e.g., the biological "mouse" vs. the computer "mouse") are present in the same visualization (e.g., a visualization of a biological mouse in the same image with a computer tower). We describe and implement a new model of creating contextual coherence in the visual imagination called Coherencer, based on the SOILIE model of imagination. We show that Coherencer is able to generate scene descriptions that are more coherent than SOILIE's original approach as well as a parallel connectionist algorithm that is considered competitive in the literature on general coherence. We also show that co-occurrence probabilities are a better association representation than holographic vectors and that better models of coherence improve the resulting output independent of the association type that is used. Theoretically, we show that Coherencer is consistent with other models of cognitive generation. In particular, Coherencer is a similar, but more cognitively plausible model than the C model of concept combination created by Costello and Keane (2000). We show that Coherencer is also consistent with both the modal schematic indices of perceptual symbol systems theory (Barsalou, 1999) and the amodal contextual constraints of Thagard's (2002) theory of coherence. Finally, we describe how Coherencer is consistent with contemporary research on the hippocampus, and we show evidence that the process of making a visualization coherent is serial.

This thesis describes an analysis of contextual coherence in the visual imagination using three disciplines: cognition, computation, and neuroscience. I examine the topic by augmenting a model of the visual imagination, SOILIE, with an improved version of their top-n model of coherence. I show that the augmented, serial local hill search model, Coherencer, is an improvement over the original model using a new, quantitative evaluation. I then demonstrate that Coherencer is better than a competitive model from the literature on general coherence; it is better than the original top-n model across different compression representations, mainly cooccurrence probabilities and holographic vectors; and it is consistent with contemporary,