Abstract
Grid cells enable efficient modeling of locations and movement through path integration. Recent work suggests that the brain might use similar mechanisms to learn the structure of objects and environments through sensorimotor processing. This work is extended in our network to support sensor orientations relative to learned allocentric object representations. The proposed mechanism enables object representations to be learned through sensorimotor sequences, and inference of these learned object representations from novel sensorimotor sequences produced by rotated objects through path integration. The model proposes that orientation-selective cells are present in each column in the neocortex, and provides a biologically plausible implementation that echoes experimental measurements and fits in with theoretical predictions of previous studies.
Highlights
We perceive the world around us through sensory experience, interpreting bottom-up sensory input with internal top-down expectations (Gilbert and Sigman, 2007)
To asses the value of the proposed network, we first evaluate its ability to effectively estimate and represent object orientations. This core computation is based on Equation (15), which was used to evaluate the orientational selectivity of the network in a similar way that head direction cells are evaluated
25 grid cell modules and 13 cells per axis gave the network enough capacity and angular resolution to work at a suitable level, and was chosen
Summary
We perceive the world around us through sensory experience, interpreting bottom-up sensory input with internal top-down expectations (Gilbert and Sigman, 2007). Saccades are an example of stable perceptions despite the continuous movement of our sensors (eyes) during sensory experience. This stable perception is invariant to the order in which the sequence of samples of the environment occurs, and is refined through movement of sensors relative to objects (as we look around). These stable representations result from correct predictions of upcoming sensory input, by including upcoming self-generated movements in conjunction with the stream of sensory inputs when calculating expectations (Killian et al, 2012, 2015). Neural representations need to take into account both object-centric ( known as allocentric) and body-centric ( known as egocentric) locations of sensors in order to accommodate the necessary invariances needed for sensory inputs (Burgess, 2006)
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
