5 measurements on self learning from rover:

6 measurements on self learning form sitter:

10 measurements on self learning from wtb:

4 measrements on long term test from rover:

(right punishment)

3 measrements on long term test from sitter (fly stops a lot during test, this result may not be accurate):

(13:left punishment, 19: right punishment)

7 measrements on long term test from wtb

(left punishment)

PKG rover self learning N=5

rover test on next day N=2

WTB self learning N=7

WTB test on next day N=4

2 rover, 4 wtb, 1 sitter test result after 24 hours

By contrast of PKG rover, this result shows PI from wtb. N = 4

test1 PI= -0.05

test2 PI= -0.08

There is the result of 20 measurements on PKG rover.

pretest:-0.45

train1:0.56

train2:0.71

test1:0.47

train3:0.83

train4:0.87

test2:0.39

After camera and punish laser adjusted, flies showed a high avoidence in the trail. Especially, when the fly was hit by punish laser in the first time of each training block. Unfortunately, there was a huge drift in the last trace after adjustment, but it shows a good avoidence in each training block.

Wild Type Drosophila were used for single fly trace testing. The basic ideal is to record the spontaneous behavior (flying trace) after punishment training by heat. The followings are 3 subjects were test with performance index (PI).

•  wtb_08 (right punishment).

• wtb_10 (left punishment).

• wtb_14 (left punishment).

• Average.

PS: From Julien’s paper, he demonstrates that there are 7 blocks were included in PI ( one pretest:PI1; four training test: PI2, PI3, PI5,PI6; two memory test: PI4,PI7)

This is the same experiment as I previously showed of Gr66a>Chrimson (ATR). The only difference is that the light was on for the whole experiment, so that the flies could see the light before the entered the arm. Previously the light switched on once the fly went into the arm. The phenotype is much stronger (there is some classical component in it). I was trying to reinforce left or right turns but it does not seem to work after a bit trying out. It makes sense ecologically I think, that the right or left turns are not coupled to the reinforcement systems. I also have been thinking about the CS-US relation bitter taste-turn directions does not make sense ecologically, but maybe if instead of bitter, I apply pain or heat …it could work. I was thinking of reinforcing orientation as well as a speed threshold, or any other variants. What do you think? I would appreciate some ideas. Since I want to make sure about what am I measuring: operant/place/classical…

linearmodelling

This is the markdown showing the protocol and results of the modelling for the choice in the T-maze. This is for calculating valence. Nevertheless, this needs to be confirmed with the results of more lines, it could be that it is overfitted, I would like to do in addition cross-validation. I´m actually doing crosses and finding new lines to have more lines to test.

So I got new flies to test. After one day of testing, I had only one control fly going through the test, all the 5 other were dead. That one did not learn.

I thus confirm that there is a problem with these flies. This means that I have no phenotype with one RNAi, and one phenotype with the other non-triggered RNAi which is probably not self-learning specific. This also means that the RNAi induction may not be sufficient and that I have no positive control for that.

The RNAi data are therefore not conclusive at all (neither positive nor negative results can be linked to any gene).

PS: I have checked the flies and overexpression of GFP did occur after 2 days Heatshock.

I will check the mutant again next week and I will be over with learning experiments. Still need some anatomy but then, I can finish the paper.

While Bjoern is in Hawaii, I used the flight simulator.

I got sufficient good flies to do 4 experiments. That’s really encouraging that the majority of the flies survived, kept their hook and flew for more than the 20 minutes required!

The other side of the coin is that I did not had any learning effect, and fly do not even avoid the punished side during the training phases: it seems something’s wrong with the laser. I bet that there is a very important trick I don’t know about the laser alignment…

but Everything will work!