Figure 4.

The general binding mechanism illustrated in figure 3 permits information about low-level visual features to be projected up to the higher layers of the network, where such information may be used by later brain systems to guide decision-making and behaviour. (a) A three-neuron binding circuit which is similar to that shown in figure 3a. However, now the binding neuron 3 is located in the higher layer along with neuron 2 representing the higher-level feature. In this case, the binding neuron 3 in the higher layer represents the presence of the lower-level feature (e.g. vertical bar) represented by neuron 1, as well as the fact that the lower-level feature is part of the higher-level feature (e.g. letter T) represented by neuron 2. Thus, information about the presence of the lower-level feature, and the fact that it is part of the higher-level feature, has been projected up to the higher network layer. (b) This process could continue up through successive network layers. Here neuron 6 is a form of binding neuron that receives inputs from neurons 3 and 5, and represents the fact that the lower-level feature (e.g. vertical bar) is part of the higher-level feature (e.g. letter T), which in turn is part of the highest level feature (e.g. word CAT). Consequently, information about the lowest level feature is projected up to the highest network layer. Reproduced with permission from Eguchi et al. [9]. Copyright © 2018 American Psychological Association. Reproduced [or Adapted] with permission. The official citation that should be used in referencing this material is [9]. No further reproduction or distribution is permitted without written permission from the American Psychological Association.