Limits...
Modeling place field activity with hierarchical slow feature analysis.

Schönfeld F, Wiskott L - Front Comput Neurosci (2015)

Bottom Line: Simulations are based on a hierarchical Slow Feature Analysis (SFA) network topped by a principal component analysis (ICA) output layer.The slowness principle is shown to account for the main findings of the presented experimental studies.Future iterations of the model will thus have to incorporate additional information, such as path integration and grid cell activity, in order to be able to also replicate studies that take place during darkness.

View Article: PubMed Central - PubMed

Affiliation: Theory of Neural Systems Group, Institut für Neuroinformatik, Ruhr Universität Bochum Bochum, Germany.

ABSTRACT
What are the computational laws of hippocampal activity? In this paper we argue for the slowness principle as a fundamental processing paradigm behind hippocampal place cell firing. We present six different studies from the experimental literature, performed with real-life rats, that we replicated in computer simulations. Each of the chosen studies allows rodents to develop stable place fields and then examines a distinct property of the established spatial encoding: adaptation to cue relocation and removal; directional dependent firing in the linear track and open field; and morphing and scaling the environment itself. Simulations are based on a hierarchical Slow Feature Analysis (SFA) network topped by a principal component analysis (ICA) output layer. The slowness principle is shown to account for the main findings of the presented experimental studies. The SFA network generates its responses using raw visual input only, which adds to its biological plausibility but requires experiments performed in light conditions. Future iterations of the model will thus have to incorporate additional information, such as path integration and grid cell activity, in order to be able to also replicate studies that take place during darkness.

No MeSH data available.


Related in: MedlinePlus

Sampling of three already established place fields over the course of an 8 min long trajectory. Despite the place fields being stable from the beginning, they appear to change position and/or split up depending on the coverage of the terrain over time. For each cell three columns are shown; from left to right these are: artificial spikes along the agent's trajectory, based on the activity values produced by our network; accumulated spikes collected in a grid of rectangular bins over the environment; the smoothed bin data as it is commonly presented in experimental papers. The activity was plotted after 30 s and 1, 2, 4, and 8 min of exploration; the number in each row denotes the peak firing activity of the corresponding cell. It can be seen that even though the place fields are stable from the beginning, measurements take time to depict stale firing fields.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4441153&req=5

Figure 4: Sampling of three already established place fields over the course of an 8 min long trajectory. Despite the place fields being stable from the beginning, they appear to change position and/or split up depending on the coverage of the terrain over time. For each cell three columns are shown; from left to right these are: artificial spikes along the agent's trajectory, based on the activity values produced by our network; accumulated spikes collected in a grid of rectangular bins over the environment; the smoothed bin data as it is commonly presented in experimental papers. The activity was plotted after 30 s and 1, 2, 4, and 8 min of exploration; the number in each row denotes the peak firing activity of the corresponding cell. It can be seen that even though the place fields are stable from the beginning, measurements take time to depict stale firing fields.

Mentions: To examine the (a) development and (b) sampling issues of place fields we present the results of two different simulations. (a) Figure 3 shows the development of three different place fields after 30 s and 1, 2, 4, and 8 min of exploration time. It can be seen that the hierarchical network produces distinct spatial fields of activity as early as within 2 min of exploration in an unknown environment. (b) Figure 4 shows the emergence of three fully established place fields while the (virtual) animal randomly traverses the environment for 8 min. Network activity is depicted after 30 s and 1, 2, 4, and 8 min of random foraging. Despite the place fields being established from the beginning, Figure 4 shows that it can take up to 8 min before their actual shape becomes clearly visible. It can also be seen that during this time the incomplete sampling leads to the place fields grow (cell 1), split and change position (cell 2), or lose alternative firing locations (cell 3). In addition, Figure 5 shows the firing pattern of a simulated cell as the arena is being traversed while the agent is consistently looking in one of eight different directions. As can be seen, place fields produced by our model fire independently of direction when being trained by random exploration in the open field.


Modeling place field activity with hierarchical slow feature analysis.

Schönfeld F, Wiskott L - Front Comput Neurosci (2015)

Sampling of three already established place fields over the course of an 8 min long trajectory. Despite the place fields being stable from the beginning, they appear to change position and/or split up depending on the coverage of the terrain over time. For each cell three columns are shown; from left to right these are: artificial spikes along the agent's trajectory, based on the activity values produced by our network; accumulated spikes collected in a grid of rectangular bins over the environment; the smoothed bin data as it is commonly presented in experimental papers. The activity was plotted after 30 s and 1, 2, 4, and 8 min of exploration; the number in each row denotes the peak firing activity of the corresponding cell. It can be seen that even though the place fields are stable from the beginning, measurements take time to depict stale firing fields.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4441153&req=5

Figure 4: Sampling of three already established place fields over the course of an 8 min long trajectory. Despite the place fields being stable from the beginning, they appear to change position and/or split up depending on the coverage of the terrain over time. For each cell three columns are shown; from left to right these are: artificial spikes along the agent's trajectory, based on the activity values produced by our network; accumulated spikes collected in a grid of rectangular bins over the environment; the smoothed bin data as it is commonly presented in experimental papers. The activity was plotted after 30 s and 1, 2, 4, and 8 min of exploration; the number in each row denotes the peak firing activity of the corresponding cell. It can be seen that even though the place fields are stable from the beginning, measurements take time to depict stale firing fields.
Mentions: To examine the (a) development and (b) sampling issues of place fields we present the results of two different simulations. (a) Figure 3 shows the development of three different place fields after 30 s and 1, 2, 4, and 8 min of exploration time. It can be seen that the hierarchical network produces distinct spatial fields of activity as early as within 2 min of exploration in an unknown environment. (b) Figure 4 shows the emergence of three fully established place fields while the (virtual) animal randomly traverses the environment for 8 min. Network activity is depicted after 30 s and 1, 2, 4, and 8 min of random foraging. Despite the place fields being established from the beginning, Figure 4 shows that it can take up to 8 min before their actual shape becomes clearly visible. It can also be seen that during this time the incomplete sampling leads to the place fields grow (cell 1), split and change position (cell 2), or lose alternative firing locations (cell 3). In addition, Figure 5 shows the firing pattern of a simulated cell as the arena is being traversed while the agent is consistently looking in one of eight different directions. As can be seen, place fields produced by our model fire independently of direction when being trained by random exploration in the open field.

Bottom Line: Simulations are based on a hierarchical Slow Feature Analysis (SFA) network topped by a principal component analysis (ICA) output layer.The slowness principle is shown to account for the main findings of the presented experimental studies.Future iterations of the model will thus have to incorporate additional information, such as path integration and grid cell activity, in order to be able to also replicate studies that take place during darkness.

View Article: PubMed Central - PubMed

Affiliation: Theory of Neural Systems Group, Institut für Neuroinformatik, Ruhr Universität Bochum Bochum, Germany.

ABSTRACT
What are the computational laws of hippocampal activity? In this paper we argue for the slowness principle as a fundamental processing paradigm behind hippocampal place cell firing. We present six different studies from the experimental literature, performed with real-life rats, that we replicated in computer simulations. Each of the chosen studies allows rodents to develop stable place fields and then examines a distinct property of the established spatial encoding: adaptation to cue relocation and removal; directional dependent firing in the linear track and open field; and morphing and scaling the environment itself. Simulations are based on a hierarchical Slow Feature Analysis (SFA) network topped by a principal component analysis (ICA) output layer. The slowness principle is shown to account for the main findings of the presented experimental studies. The SFA network generates its responses using raw visual input only, which adds to its biological plausibility but requires experiments performed in light conditions. Future iterations of the model will thus have to incorporate additional information, such as path integration and grid cell activity, in order to be able to also replicate studies that take place during darkness.

No MeSH data available.


Related in: MedlinePlus