Outline:
functional classification for the visual system:
Cluster 10 and cluster 11 (Fig. 2c) both receive strong input from photoreceptors R1–6 (Extended Data Fig. 11), and we propose that they are regarded as OFF and ON channels, respectively, carrying information about light decrements (OFF stimuli) and light increments (ON stimuli).
From: Neuronal parts list and wiring diagram for a visual system
OFF cells: L1, L2, L4, Tm1, Tm2 and Tm4, C2 and C3 ON cells: L5, Mi1 and Tm3
The ON and OFF motion pathways were traditionally defined by working backwards from the T4 and T5 motion detectors
(Cluster 10 contains the OFF cells L2, L4, Tm1, Tm2 and Tm4. Cluster 11 contains the ON cells L5, Mi1 and Tm3, and also the OFF cell L1. It makes sense to assign L1 to the ON channel even though it is an OFF cell, because L1 is inhibitory/glutamatergic, so its effects on downstream partners are similar to those of an ON excitatory cell.)
Luminance channel: L3 constitutes a separate luminance channel. L3 is the only L type with a sustained rather than transient response, and it encodes luminance rather than contrast. Dm4, Dm9, Dm12, Dm20 and Mi9, which all have L3 as their strongest input.
Feedback: Lai, Lawf1, Lai and Lawf2
provides centrifugal feedback to the OFF channel
From: Neuronal parts list and wiring diagram for a visual system
The motion-detecting： T4 and T5 families
CT1 and Tm9 are well known to be important for motion computation
It makes sense to regard Tm9 as dedicated to the motion subsystem rather than part of a general-purpose OFF channel, as 80% of its output synapses are onto CT1 or T5.
Cluster 16 also includes Li14, an interneuron type with T5a as the strongest input, and T5a through T5d as the strongest outputs.
more about T4a-T4d, T5a-T5d neurons
T2 and T3, which have been implicated in the detection of small objects
Their downstream VPN partners LC11 and LC18 (Fig. 6) are also activated by small objects.
The inner photoreceptors R7 and R8 are important for Drosophila colour vision because their responses are more narrowly tuned to the wavelength of light
Dm8a, Dm8b, Dm11 and DmDRA2, which are all inner photoreceptor targets
From: Color vision in insects: insights from Drosophila
Locus coeruleus
1. natural stimulus (female movement)
From: Mapping model units to visual neurons reveals population code for social behaviour
Real LC responses and predicted responses to stimulus sequences of a moving fictive female:
Visual features of female motion are distributed across the population of model LC units:
The role of LC neurons in the sensorimotor transformation of the male fly during courtship(a summary):
they also provide the ephys data of LC neuron, which can be used to compare with the model prediction:
(LC4, LC6, LC9, LC10ad, LC10a, LC10bc, LC10d, LC11, LC12, LC13, LC15, LC16, LC17, LC18, LC20, LC21, LC22, LC24, LC26, LC31)
https://dandiarchive.org/dandiset/000951/0.240418.2218/files?location=&page=1
https://neurosift.app/dandiset/000951?dandisetVersion=0.240418.2218
2. artificial stimulus
From: A functionally ordered visual feature map in the Drosophila brain
there are ∼20 anatomically distinct LC types (including LC-like visual projection cell types called lobula plate lobula columnar, or LPLC, neurons, each comprising a population of between 30 and 220 neurons)
We discover novel detectors for objects smaller than the lens resolution (LC18) and for complex line motion (LC25). each cell integrates visual inputs within a compact region and is likely to act as a local feature detector
several reason to use the the component of the neuron instead of the point neuron:
the connection between the point neurons in more like full connection, compared to the component neuron, which only have less than ~5 connected component
evidence from paper: connectom constrain neural networks
Connectome measurements constrain neural networks in circuits with sparse connectivity:
https://www.nature.com/articles/s41586-024-07939-3/figures/5
in that case, the sparse connection will have a more similar parameter between the student network and the teacher network compared to the dense connection
morphology will influence the function of the neuron:
from: How fly neurons compute the direction of visual motion
https://www.cell.com/current-biology/fulltext/S0960-9822(22)01023-5
https://www.sciencedirect.com/science/article/pii/S0896627322001787 there are ∼20 anatomically distinct LC types (including LC-like visual projection cell types called lobula plate lobula columnar, or LPLC, neurons, each comprising a population of between 30 and 220 neurons)
We discover novel detectors for objects smaller than the lens resolution (LC18) and for complex line motion (LC25). each cell integrates visual inputs within a compact region and is likely to act as a local feature detector
https://ars.els-cdn.com/content/image/1-s2.0-S0896627322001787-gr2_lrg.jpg
LC neurons are tuned to the size of visual objects
Detecting object size by comparing contrast changes in time (??how to compare the contrast change in time? there might be a time-sensitive mechanism, like the delay between the synapse? the ion channel? the distance between different compoenent?)
https://www.nature.com/articles/s41586-025-08746-0
https://www.nature.com/articles/s41586-025-08746-0/figures/6 all most all the projection neuron have to go through a bottleneck we can use this part to split the neuron we want to study
https://www.nature.com/articles/s41586-024-07953-5 H. Sebastian Seung Nature 2024
Dm3v, Dm3p, Dm3q, TmY4, TmY9q, TmY9q^{⟂} check the recpetive field of those neuron, by giving them different visual input
https://elifesciences.org/articles/21022 2016
activate LC neuron -> movement forward walking, jumping, reaching, backward walking and turning,
https://www.nature.com/articles/s41586-024-07451-8 2024
using female as different visual input to test the LC neuron
female movement: size, position, rotation give use more complex visual input feature and related reaction of LC neuron
fly-119.png
question: they recored the LC-silenced behavior, do they have the activity recording of the LC neuron? Fig 2
Fig2 f: Different model LC units appear either to directly encode a visual parameter (for example, LC10a encodes size) or encode features derived from the parameter, such as a delay (LC17, arrows) or speed at which female size changes (LC13).
data: only model are publishded, the lc neuron activity not
