This poster describes a two-layer network model that uses cortical grid cells and path integration to learn and recognize objects through movement. In our model, one layer contains several grid cell-like modules and provides a location signal for each learned object such that features can be associated with a specific location in the reference frame of that object. A second layer, a sensory input layer, receives the location representation as context, and uses it to encode the sensory input in the context of a location in the object’s reference frame.
At this point in time there is no consensus in neuroscience literature on how grid cells are involved in the representation of 3D location, and their contribution to coding variables beyond 2 or 3 dimensions is completely uncharted territory. This poster explores how grid cells can encode N-dimensional variables, using random velocity projections. The poster covers path integration, relation to band cells, and capacity and tuning curve.
This poster introduces a proposal that the brain uses grid cells to perform unsupervised learning of landmark locations. It shows the results of a network model trained on 1000 environments, compared to a bag-of-features model. It also lays out discussion topics for future extensions of this work.
This poster highlights one of the foundational topics of Numenta research: sparse distributed representations, or SDRs for short. SDRs are how the brain represents information. The mathematical properties of SDRs are essential components of biological intelligence. This poster examines how accurately neurons can recognize sparse patterns.
In this poster, we show how the brain might use a grid cell code to represent 1) sensed structures at locations in viewer-centric coordinates and 2) sensed features and locations in object-centric coordinates. We lay out a mechanism that shows the transform routes between grid cells that enable object recognition.