Figure 1.
(A) 16 glomerulus connected through SAs. The thick red line represents full connectivity between glomerulus. b) Architecture of the glomerular unit. The cells depicted are: mitral cell (MC), external tufted cell (ET), periglomerular cell (PG), and short axon (SA). Black balls represent inhibitory synapses and red balls indicate excitatory synapses. The ORN input synapses into MC, ET and PG cells. MC and PG cells form a negative feedback loop, where PG cells inhibit MC cells and in turn MC excite PG cells. ET cells contribute to the inhibition of MC through an excitatory connection to PG cells. Finally, connections between glomerulus are achieved via SA cells, which receive excitatory inputs from ET cells and transmit its outputs to PG and ET in an excitatory fashion.
Figure 2.
Input odor patters and MC output activity.
(A) Scores plot of the first two principal components of the input odors used to analyze the glomerular layer model. The directions of the arrows indicate increasing concentration. (B) Relative composition of odor C and E on the series of binary mixtures that simulate the morphing between the two odors. The y-axis shows the relative composition of odor C and odor E in each one of the 21 mixtures. (C) Example of mitral cell responses of a 16-glomeruli model. Different colors identify different mitral cells.
Table 1.
Model parameters of the olfactory bulb neurons.
Figure 3.
MCs and ET cells output in the 6 odors - 6 concentrations experiment.
(A) Scores plot of the first three principal components of the MCs output. The output of MC is obtained as the mean firing rate during exposure of 0.5 s to the odors. (B) Scores plot of the first two principal components of the ET cells output. The output of ET cells is also obtained as the mean firing rate during the same experiment. Arrows indicate increasing concentration. (C) Scores of the first principal component of the ET output versus the stimulus concentration.
Figure 4.
Quantitative measure of odor identity and concentration information in the 6 odors - 6 concentrations experiment.
(A) Fisher's discriminant ratio of MC outputs, ETs cells outputs and input odor patterns in the 6 odors - 6 concentrations experiment. It is computed for different values of the short axon cell synaptic weights, which regulates the connection strength between glomeruli. (B) Pearson's correlation coefficient between MC, ET outputs, input odor patterns and the input odor concentrations for different short axon cell synaptic weights. The error bars show the standard deviation of the Pearson's correlation coefficient across different odors.
Figure 5.
Hierarchical clustering of input patterns in the odor morphing experiment.
We performed hierarchical clustering based on k-means on a sequence of binary mixtures going from odor C to odor E through intermediate mixtures (morphing) along with the odor patterns of the 6 odors - 6 concentrations experiment. Clustering results are presented as a dendogram in terms of the distance to the k-means nearest cluster. Odors are identified by color, where high concentrations are plotted in a solid line and low concentrations are plotted in a dashed line.
Figure 6.
Hierarchical clustering of MC outputs in the odor morphing experiment.
In this hierarchical clustering, MC outputs are grouped according to their identity in the case of pure odors. Mixtures cluster together and also with the different concentrations of pure odors C and E, which are the two components of the mixtures.
Figure 7.
Hierarchical clustering of ET cells output in the odor morphing experiment.
The hierarchical clustering of ET output patterns show a clear separation between high concentration odors (red lines) and low concentration odors (blue lines). Mixture odors lie within the high concentration cluster but not far from low concentrations. This is consistent with a proper disposition of concentrations since mixtures are formed by two components of concentration factors of 1 multiplied by mixing factor that sum up to 1 in all cases.
Figure 8.
Smooth evolution of mixture odors in the morphing experiment.
The figure represents the score plots of the 3 principal components of the MCs outputs in the morphing experiment. The 21 mixtures evolve smoothly from the initial odor C to the final odor E.