Limits...
Noise promotes independent control of gamma oscillations and grid firing within recurrent attractor networks.

Solanka L, van Rossum MC, Nolan MF - Elife (2015)

Bottom Line: Neural computations underlying cognitive functions require calibration of the strength of excitatory and inhibitory synaptic connections and are associated with modulation of gamma frequency oscillations in network activity.This beneficial role for noise results from disruption of epileptic-like network states.Our results have implications for tuning of normal circuit function and for disorders associated with changes in gamma oscillations and synaptic strength.

View Article: PubMed Central - PubMed

Affiliation: Centre for Integrative Physiology, University of Edinburgh, Edinburgh, United Kingdom.

ABSTRACT
Neural computations underlying cognitive functions require calibration of the strength of excitatory and inhibitory synaptic connections and are associated with modulation of gamma frequency oscillations in network activity. However, principles relating gamma oscillations, synaptic strength and circuit computations are unclear. We address this in attractor network models that account for grid firing and theta-nested gamma oscillations in the medial entorhinal cortex. We show that moderate intrinsic noise massively increases the range of synaptic strengths supporting gamma oscillations and grid computation. With moderate noise, variation in excitatory or inhibitory synaptic strength tunes the amplitude and frequency of gamma activity without disrupting grid firing. This beneficial role for noise results from disruption of epileptic-like network states. Thus, moderate noise promotes independent control of multiplexed firing rate- and gamma-based computational mechanisms. Our results have implications for tuning of normal circuit function and for disorders associated with changes in gamma oscillations and synaptic strength.

No MeSH data available.


Gridness scores of I cells.Colour plots show gridness score as a function of gE and gI for networks without noise (A), with noise standard deviation σ = 150 pA (B), and σ = 300 pA (C). Data are from simulations of networks with feedback inhibition only (E-I networks; Figure 2). Black lines indicate the region from Figure 2D–F where the gridness score of E cells = 0.5.DOI:http://dx.doi.org/10.7554/eLife.06444.008
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4508578&req=5

fig2s3: Gridness scores of I cells.Colour plots show gridness score as a function of gE and gI for networks without noise (A), with noise standard deviation σ = 150 pA (B), and σ = 300 pA (C). Data are from simulations of networks with feedback inhibition only (E-I networks; Figure 2). Black lines indicate the region from Figure 2D–F where the gridness score of E cells = 0.5.DOI:http://dx.doi.org/10.7554/eLife.06444.008

Mentions: How closely does the firing of I cells in the simulated networks correspond to inhibitory activity in behaving animals, and to what extent is the pattern of I cell firing affected by gE, gI and noise? While there is little data on the spatial firing of interneurons in the MEC, recent evidence indicates that the majority of parvalbumin positive interneurons have firing fields with significant spatial stability, but low spatial sparsity and grid scores compared to excitatory grid cells (Buetfering et al., 2014). A possible interpretation of these data is that parvalbumin positive cells are unlikely to fulfill the roles of I cells predicted in E-I models. However, in networks that we evaluate here in which E cells have grid firing fields in the presence of moderate noise, I cell firing fields also have a much lower spatial information content and spatial sparsity than the corresponding E cell firing fields (E cells: spatial sparsity 0.788 ± 0.061, spatial information: 1.749 ± 0.32 bits/spike; I cells: spatial sparsity 0.239 ± 0.018, spatial information 0.243 ± 0.024 bits/spike; p < 10−16 for comparisons of both spatial sparsity and information; paired t-test; data range is indicated as mean ± standard deviation) (Figure 2A–C and Figure 2—figure supplement 2). Spatial autocorrelograms of simulated I cell firing fields also do not contain the six hexagonally organized peaks that are characteristic of grid fields (Figure 2A–C). Nevertheless, I cell spatial autocorrelograms produce positive grid scores (0.39 ± 0.16; Figure 2—figure supplement 3), although these are reduced compared to scores for the E cells in the same networks (E cells: 0.796 ± 0.157; p < 10−16; paired t-test; mean ± SD) and in many networks are below the threshold considered previously to qualify as grid like (cf. Figure 4B of Buetfering et al., 2014). When we evaluated the dependence of I cell spatial firing on gE, gI and noise, it appeared to be similar to that of E cells (Figure 2—figure supplement 3). To assess whether grid scores of I cells can be reduced further in E-I networks while maintaining grid firing by E cells, we investigated networks in which uncorrelated spatial input is applied to each I cell (Figure 2—figure supplement 4). In these simulations E cells had grid scores of 0.57 ± 0.25, spatial sparsity of 0.78 ± 0.03 and spatial information of 1.69 ± 0.18 bits/spike, whereas I cells had grid scores of 0.16 ± 0.2 (p < 10−16, paired t-test), spatial sparsity of 0.21 ± 0.01 (p < 10−16, paired t-test) and spatial information of 0.2 ± 0.01 bits/spike (p < 10−16, paired t-test; range of all data sets is mean ± SD). Thus, spatial firing of I cells has a similar dependence on noise, gE and gI to grid cells, conventional indices of spatial firing are nevertheless much lower for I cells in E-I networks compared to E cells, and grid firing by E cells in E-I networks is relatively robust to disruption of the rotational symmetry of I cell firing fields.


Noise promotes independent control of gamma oscillations and grid firing within recurrent attractor networks.

Solanka L, van Rossum MC, Nolan MF - Elife (2015)

Gridness scores of I cells.Colour plots show gridness score as a function of gE and gI for networks without noise (A), with noise standard deviation σ = 150 pA (B), and σ = 300 pA (C). Data are from simulations of networks with feedback inhibition only (E-I networks; Figure 2). Black lines indicate the region from Figure 2D–F where the gridness score of E cells = 0.5.DOI:http://dx.doi.org/10.7554/eLife.06444.008
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4508578&req=5

fig2s3: Gridness scores of I cells.Colour plots show gridness score as a function of gE and gI for networks without noise (A), with noise standard deviation σ = 150 pA (B), and σ = 300 pA (C). Data are from simulations of networks with feedback inhibition only (E-I networks; Figure 2). Black lines indicate the region from Figure 2D–F where the gridness score of E cells = 0.5.DOI:http://dx.doi.org/10.7554/eLife.06444.008
Mentions: How closely does the firing of I cells in the simulated networks correspond to inhibitory activity in behaving animals, and to what extent is the pattern of I cell firing affected by gE, gI and noise? While there is little data on the spatial firing of interneurons in the MEC, recent evidence indicates that the majority of parvalbumin positive interneurons have firing fields with significant spatial stability, but low spatial sparsity and grid scores compared to excitatory grid cells (Buetfering et al., 2014). A possible interpretation of these data is that parvalbumin positive cells are unlikely to fulfill the roles of I cells predicted in E-I models. However, in networks that we evaluate here in which E cells have grid firing fields in the presence of moderate noise, I cell firing fields also have a much lower spatial information content and spatial sparsity than the corresponding E cell firing fields (E cells: spatial sparsity 0.788 ± 0.061, spatial information: 1.749 ± 0.32 bits/spike; I cells: spatial sparsity 0.239 ± 0.018, spatial information 0.243 ± 0.024 bits/spike; p < 10−16 for comparisons of both spatial sparsity and information; paired t-test; data range is indicated as mean ± standard deviation) (Figure 2A–C and Figure 2—figure supplement 2). Spatial autocorrelograms of simulated I cell firing fields also do not contain the six hexagonally organized peaks that are characteristic of grid fields (Figure 2A–C). Nevertheless, I cell spatial autocorrelograms produce positive grid scores (0.39 ± 0.16; Figure 2—figure supplement 3), although these are reduced compared to scores for the E cells in the same networks (E cells: 0.796 ± 0.157; p < 10−16; paired t-test; mean ± SD) and in many networks are below the threshold considered previously to qualify as grid like (cf. Figure 4B of Buetfering et al., 2014). When we evaluated the dependence of I cell spatial firing on gE, gI and noise, it appeared to be similar to that of E cells (Figure 2—figure supplement 3). To assess whether grid scores of I cells can be reduced further in E-I networks while maintaining grid firing by E cells, we investigated networks in which uncorrelated spatial input is applied to each I cell (Figure 2—figure supplement 4). In these simulations E cells had grid scores of 0.57 ± 0.25, spatial sparsity of 0.78 ± 0.03 and spatial information of 1.69 ± 0.18 bits/spike, whereas I cells had grid scores of 0.16 ± 0.2 (p < 10−16, paired t-test), spatial sparsity of 0.21 ± 0.01 (p < 10−16, paired t-test) and spatial information of 0.2 ± 0.01 bits/spike (p < 10−16, paired t-test; range of all data sets is mean ± SD). Thus, spatial firing of I cells has a similar dependence on noise, gE and gI to grid cells, conventional indices of spatial firing are nevertheless much lower for I cells in E-I networks compared to E cells, and grid firing by E cells in E-I networks is relatively robust to disruption of the rotational symmetry of I cell firing fields.

Bottom Line: Neural computations underlying cognitive functions require calibration of the strength of excitatory and inhibitory synaptic connections and are associated with modulation of gamma frequency oscillations in network activity.This beneficial role for noise results from disruption of epileptic-like network states.Our results have implications for tuning of normal circuit function and for disorders associated with changes in gamma oscillations and synaptic strength.

View Article: PubMed Central - PubMed

Affiliation: Centre for Integrative Physiology, University of Edinburgh, Edinburgh, United Kingdom.

ABSTRACT
Neural computations underlying cognitive functions require calibration of the strength of excitatory and inhibitory synaptic connections and are associated with modulation of gamma frequency oscillations in network activity. However, principles relating gamma oscillations, synaptic strength and circuit computations are unclear. We address this in attractor network models that account for grid firing and theta-nested gamma oscillations in the medial entorhinal cortex. We show that moderate intrinsic noise massively increases the range of synaptic strengths supporting gamma oscillations and grid computation. With moderate noise, variation in excitatory or inhibitory synaptic strength tunes the amplitude and frequency of gamma activity without disrupting grid firing. This beneficial role for noise results from disruption of epileptic-like network states. Thus, moderate noise promotes independent control of multiplexed firing rate- and gamma-based computational mechanisms. Our results have implications for tuning of normal circuit function and for disorders associated with changes in gamma oscillations and synaptic strength.

No MeSH data available.