Code for the research internship project with the title "Orientation selectivity can arise from distinct connectivity patterns".
Model of a single orientation-selective neuron receiving inputs from afferents with different preferred orientations. The project's aim is to investigate what properties the postsynaptic neuron has and how they depend on the input afferents. It aims to investigate two types of connectivity of neurons in the primary visual cortex, i.e., weight-based and number-based.
For a demonstration of how to instantiate simple objects using the classes below, see the jupyter notebook demo, or their respective source files in src/.
-
ConnectedNeuron.py - Class containing the stimulus-independent properties of the single neuron and its afferents, such as nr of afferents, their synaptic weights, and the individual tuning of the afferents (distribution of PO per afferent).
-
SpikyLIF.py - Class describing the activity of the postsynaptic neuron given the circuit connectivity and spiking activity of the presynaptic neurons.
-
TunedResponse.py - Class containing orientation-selective response of pre- and post-synaptic neurons upon stimuli in form of bars of different orientation angles.
To reproduce some of the results using the functional structure, the lines below can be run in the main.py file.
get_response_for_bar(trials=5, to_plot=True, mean = np.pi/4)
check_robustness()
Cristina Melnic, Douglas Feitosa Tomé, Tim P Vogels
Neurons in the primary visual cortex (V1) are known to selectively respond to stimuli of spatial orientations, whereby groups of neurons are tuned to distinct preferred orientation angles. It was established that neurons with similar preferred orientation have a higher probability of being connected, which was interpreted in classical theories as larger synaptic weights. A recent experimental study of single neurons with all their afferents [1], however, found the individual synaptic strength of afferents to be independent of the similarity to the postsynaptic preferred orientation.
Here, we theoretically investigate a neuron model with orientation selective afferents in the two scenarios of correlated and independent individual synaptic strength with respect to preferred orientation. We observe that in both cases the shape of receptive fields arises from the cumulative synaptic weight of the active afferents, independently of the specific connectivity. The model is thus consistent with findings of higher probability between neurons of similar functionality and offers a mathematical framework for further investigations of factors that give rise to orientation selectivity.
[1] Scholl et al. 2021, “Cortical response selectivity derives from strength in numbers of synapses”, Nature
This project was realised during a summer internship (June - August 2022) in the Vogels lab at IST Austria, funded by OeAD and ISTA as a part of the "ISTernship" program. It was closely supervised by Dr. Douglas Feitosa Tomé and Prof. Dr. Tim P Vogels and other members of the "Computational Neuroscience and Neurotheory" group at ISTA.
