Computational and Systems Biology

We analyse experimental data on attention to indicate that any attention feedback control signals to lower order cortical sites will lead to a quadratic sigma-pi form of output in its dependence on the lower-order input and the feedback signal. The manner by which this structure works is shown by a brief simulation. We then discuss how such a structure could arise from the action of diffuse acetylcholine signals from the NBM, especially involving nicotinic receptors. We deduce certain structural regularities which should be expected both at localand at micro-circuit level, mainly in cortical layer V (the output layer).

We investigate three possible methods of specifying the microstructure of attention feedback: contrast gain, additive and output gain, using simple single node and 3-layer cortical models composed of graded or spiking neurons. Contrast gain and additive attention are also tested in a spiking network which is simplified by mean field methods. The simulation task uses two stimuli, probe and reference, presented singly or together within the neuronal receptive fields whilst attention is directed towards or away from the receptive field. Model neurons are differentially activated in the different stimuli and attention and equilibrium potentials or average firing rates recorded depending on neuron type are recorded. We compare results for the different modes of attention and architectures with experimental single cell recordings which show how neuronal firing rates change in response to attention, with a bias towards neurons that respond more effectively to the attended stimulus, to investigate which attentional method best fits the experimental data. The simulation results are also mathematically analysed. We conclude that there is most experimental support for contrast gain, although some additional feedback gain would be possible. We propose a tentative method by which attention as contrast gain may occur in the primate brain using acetylcholine and nicotinic receptors.

Occlusion is currently at the centre of analysis in machine vision. We present an approach to it that uses attention feedback to an occluded object to obtain its correct recognition. Various simulations are performed using a hierarchical visual attention feedback system, based on contrast gain (which we discuss as to its relation to possible hallucinations that could be caused by feedback). We then discuss implications of our results for object representations per se. N.R. Taylor et al.

We propose a general attention-based approach to thinking and cognition (more specifically reasoning and planning) in cognitive machines as based on the ability to manipulate neural activity in a virtual manner so as to achieve certain goals; this can then lead to decisions to make movements or to no actions whatever. The basic components are proposed to consist of forward/inverse model motor control pairs in an attention-control architecture, in which buffers are used to achieve sequencing by recurrence of virtual actions and attended states. How this model can apply to various reasoning paradigm will be described and first simulations presented using a virtual robot environment.

We present results from a paradigm of infant motor skill learning, involving selecting the correct approach to grasp a ball placed on a low stability stand. We also construct a theoretical model, based on the concept of observational learning as the discovery of non-perceived affordances and the transfer of actions on affordances, which can explain some of the experimental data presented. We discuss the underlying concepts of the model (including its basis in neuroanatomy) and how they relate to the paradigm, and suggest possible model extensions.

We develop a general neural network-based architecture for the process of mental simulation, initially treated at a somewhat abstract level. To develop the theory further it is shown how the theory can handle observational learning as a specific form of mental simulation: simulations are presented of simple paradigms and results obtained on children undergoing tests on observational learning. Questions of learning and other aspects are treated in a discussion section.

Cognition is a complex and wide-ranging field of study. We suggest a possible approach to modelling reasoning as a particular aspect of cognition involving the use of forward and inverse internal models to allow the effect of actions on the world to be considered. The field of animal reasoning provides an excellent source of manageable problems. We describe one of these, the two-sticks paradigm, in detail and construct a model of how the reasoning process involved in solving this paradigm might operate. Results from this simulation are presented, and we discuss important features of the model, in particular the use of spatially invariant goal representations and fast movements of reward.

The hippocampus is a part of the brain strongly linked to spatial exploration. Within it exist 'place cells' which fire preferentially when an animal is in certain regions of physical space. Recent research has shown that these place cells and their corresponding representations of space behave differently in the CA3 and CA1 subfields of the hippocampus.

Spatial navagation is a critical part of animal behaviour. Experimental data show that some cells in the hippocampus of animals engaged in exploration respond preferentially to particular physical locations. These place cells give us an important indication of hippocampal participation in spatial processing. Recent work has examined differences in place field representations between hippocampal subfields. We discuss these findings and show, using a computational model, how known aspects of hippocampal physiology can explain these differences.

We discuss evidence for the existence of mirror systems in the brain, including recent experimental results that demonstrate the use of shared pathways for the observation and execution of reaching and grasping actions. We then describe a brain based model of observational learning that explains the similarities and differences in levels of activation of brain regions during observation and execution of actions. We simulate a very simple paradigm whereby an actor performs an action which is observed and then repeated by the simulated animal. We discuss the implications and possible extensions of our model.

Gibbens (2008) presents a methodology for assisting humans with analysing large amounts of image data. Split into four major sections, namely pre-processing, segmentation, region modelling and taxonomy creation ), each one can be implemented individually. With some success Gibbens demonstrates the methodology using several segmentation techniques, such as scale-spaces with blob detection and manual hand marked regions with expert consensus, along with a modified Texton approach for region modelling. This paper reviews the techniques employed by Gibbens, highlighting where improvements can be made as well as presenting the work conducted so far and discussing the development over the next two years.

With the advent of many Terrabyte size image datasets of the Moon, it is no longer possible for the lunar science community to manually count and classify all the available lunar craters. We discuss the use of automated feature recogniton and textural analysis techniques to assist in a solution to this problem.

This paper presents a simple technique to segment imagery from the Context Camera (CTX) aboard the Mars Reconnaisssance Orbiter (MRO). Using a technique known as Texton analysis, features are highlighted based upon texture. This technique is susceptible to variations in view and illumination (phase) angle however provides an effective method of segmentation.

Automated methods for dealing with large amounts of image data can save researchers a great deal of time. However computers often have trouble comparing similar images viewed from different view and illumination conditions (phase angle). The work presented in this poster demonstrates a method to automatically classify and visualise large amounts of image data as well as providing a solution to issues of view and illumination angles when viewing remotely sensed imagery.

Gibbens (2008) presents a methodology for assisting humans with analysing large amounts of image data. Split into four major sections, namely pre-processing, segmentation, region modelling and taxonomy creation ), each one can be implemented individually. With some success Gibbens demonstrates the methodology using several segmentation techniques, such as scale-spaces with blob detection and manual hand marked regions with expert consensus, along with a modified Texton approach for region modelling. This paper reviews the techniques employed by Gibbens, highlighting where improvements can be made as well as presenting the work conducted so far and discussing the development over the next two years.

Image processing and classification is an important part of analysis in many fields of scientific research. This poses a problem in that humans are unable to analyse large data sets quickly or accurately . Gibbens [2] presents a method of image set analysis where similarity can be described by distance. The proposed framework for his methodology involved segmenting an image either automatically or by hand to form regions. Image region modelling was then performed to produce a tractable description of the region. This description was used to compare other regions and create a taxonomy. A visual representation of similar regions was generated (see ) using a clustering algorithm. Subsequently, a taxonomy was created from the NASA Clementine spacecraft [3] data which classified craters on the Moon (see ) automatically. These classifications were compared to the Baldwin classifier [4] to some success.