Kenneth D Harris
@kennethd_harris
Followers
6K
Following
2K
Media
47
Statuses
729
Neuroscientist at University College London
London, England
Joined April 2016
RT @ObhishekSaha: 🚨Open letter calling upon the UK government to implement the Freedom of Speech Act in full🚨. If you are a UK-based academ….
0
87
0
RT @EnnyvBeest: [1/5] Our paper “Tracking neurons across days with high-density probes” is now out in @naturemethods! .
0
69
0
RT @EnnyvBeest: The 2024 UCL Neuropixels course will take place on 14-16 October. Registration is free. Register today! More info: https://….
0
37
0
10. So in this task, MOs encodes a perseverative decision variable, and is required for the updating of this variable and its effect on choices and reaction time. mPFC is more involved in reward-based learning.
0
0
3
9. Optogenetic suppression of MOs during the choice period reduced perseverative behavior and perseverative learning, and increased reaction times. In contrast, suppressing mPFC following reward delivery slowed reward-based learning.
1
0
6
8. Statistical methods robust to nonsense correlations showed that while mice waited for the cue, the perseveration variable was encoded specifically in MOs; the reward-seeking variable was encoded nowhere. Once movements started, both variables were encoded in many regions.
1
0
0
7. Statistical analysis of this kind of data is hard: applying standard tests can lead to “nonsense correlations” because of slow drifts in behavior sequences and neural recordings.
1. In neuroscience we often analyze variables that change slowly over time, for which standard statistical methods can give bogus results. I explored ten methods to deal with this problem and wrote up the results in a preprint.
1
0
0
6. But these neurons didn’t just encode the upcoming choice. They fired more after long sequences of the same choice, correlating with the perseveration variable of a cognitive models fit to their behavior. They also predicted rapid reaction times.
1
0
0
5. Neuropixels recordings across the frontal cortex showed correlates of upcoming choices when waiting for the go cue, almost exclusively in dorsal prefrontal cortex (area MOs).
1
0
0
4. We trained mice to perform a dynamic reversal learning task. Their behavior could be fit well by cognitive models, but only models including perseveration. Mice responded faster when perseverating, but got less reward than if they had made optimal choices.
1
0
0
3. By that definition, all humans and all animals are insane. Because we perseverate: we repeat actions even if not rewarded. The neural basis of perseveration has seen little attention, especially compared to reward-based learning.
1
0
1
2. “Insanity is doing the same thing over and over again and expecting different results.”.Sometimes attributed to Einstein, this quote seems to be first from a visitor to Al-Anon, Knoxville TN, in 1981.
1
0
1
1. New preprint on the neural basis of perseverative behavior. With @neurvanna, Yeqing Wang, Laura Funnell, Bex Terry, @JinOh555, and @kevinjmiller10.
3
13
70
RT @Michael_J_Black: Young scientists regularly ask me for career advice. Academia or industry? Big company or startup? US or Europe? Good….
0
241
0
Are you interested in how neuronal populations integrate external and internal signals to drive behavior? Join our lab as a postdoc! . We are looking for people with diverse skills, from computational to experimental.For more information, see
0
33
87
RT @michael_okun: Exciting preprint of @jasmanley et al. (Alipasha Vaziri's lab) extending the findings of @computingnature, @marius10p et….
0
6
0
New paper from Matteo Carandini. Because he is off twitter, I am posting it for him!
1
13
100
8. Conditioning on only past events is called tangent randomization. Unlike conditional randomization, tangent randomization is not always valid. There’s an example in the paper. For more details on when and why tangent randomization works see
0
0
2