OSCR

Distinct Disinhibitory Circuits Link Short-Term Adaptation to Familiarity and Reward Learning in Visual Cortex

Code ↔ Paper

The paper beside its authors' code: matches between them have not been computed for this paper yet.

Paper

Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Jupyter notebook · 642 lines · 22 KB · Apache-2.0

  1. # %% [markdown]
  2. # # Model Implementation Habituation Session 1
  4. # %% [markdown]
  5. # # Imports
  7. # %%
  8. import numpy as np
  9. import random
  10. import pandas as pd
  11. import matplotlib.pyplot as plt
  12. from scipy.integrate import odeint, solve_ivp
  13. import seaborn as sns
  14. from sklearn.metrics import mean_squared_error, r2_score
  15. from lmfit import Parameters
  16. from IPython import display
  18. # %% [markdown]
  19. # # Model diagram
  21. # %%
  22. display.Image(r'Model_diagram.png')
  24. # %% [markdown]
  25. # # Functions
  27. # %%
  28. def Step(t, stim = 5, inter = 10, amp = 1.0, base = 0, k = 0, delay = 0, decay_f = 0):
  29. """
  30. Feedback (FB) input:
  31. ------------------------------------
  32. Represented as a step function. Returns value of the input function in a time point t.
  33. ------------------------------------
  34. Parameters:
  36. t (running) - time in seconds;
  37. stim (fixed) = 5 s - time of the stimulus start;
  38. inter (fixed) = 10 s - duration of the stimulus;
  39. amp (fixed) = 1.0 Hz - amplitude of response;
  40. base (fixed) = 0 Hz - baseline activity;
  41. k (fixed) = 0 - slope of the slow component. Used to be varied while testing linear depression or sensitization component in the FB input;
  42. delay (variable) - delay of the FB input to the cell relative to the stimulus start;
  44. """
  46. if (t < stim + delay):
  47. h = base
  48. elif (t > stim + inter):
  49. h = amp * np.exp(-(t - stim - inter)*decay_f)
  50. else:
  51. h = amp*(t - stim - delay)*k*0.164745 + amp
  53. return h
  55. # %%
  56. def Sigm(t, stim = 5, inter = 10, ampl = 1.0, base = 0, rate = 1, delay = 0, decay_s = 1, dt = 0):
  57. """
  58. Slow Sensitizing (SS) input:
  59. ------------------------------------
  60. Represented as a sigmoid function. Returns value of the input function in a time point t.
  61. ------------------------------------
  62. Parameters:
  64. t (running) - time in seconds;
  65. stim (fixed) = 5 s - time of the stimulus start;
  66. inter (fixed) = 10 s - duration of the stimulus;
  67. ampl (fixed) = 1.0 Hz - amplitude of response;
  68. base (fixed) = 0 Hz - baseline activity;
  69. rate (variable) - time-constant of the SS input;
  70. delay (variable) - shift of the sigmoid center relative to stimulus start;
  71. decay_s (variable) - time-constant of the SS input exponential decay after the end of stimulation;
  73. """
  75. if (t < stim):
  76. h = base
  77. elif (t > stim + inter):
  78. h = (base + (ampl/(1 + np.exp((delay-inter - dt)/rate))))*np.exp(-(t - stim - inter)*decay_s) + base
  80. else:
  81. h = base + (ampl/(1 + np.exp((stim + delay - t)/rate))) #Actualy rate here is time constant, where 1/rate is actual rate
  84. return h
  86. # %%
  87. def expon(t, stim = 5, inter = 10, ampl = 1.5, base = 0, decay = 1, delay = 0, b = 0, decay_ff = 0, s_start = 0.1, k = 0.1):
  88. """
  89. Feedforward (FF) input:
  90. ------------------------------------
  91. Represented as a flat step function with fast exponential decay on the stimulus start and linear increase during 10 second period.
  92. Returns value of the input function in a time point t.
  93. ------------------------------------
  94. Parameters:
  96. t (running) - time in seconds;
  97. stim (fixed) = 5 s - time of the stimulus start;
  98. inter (fixed) = 10 s - duration of the stimulus;
  99. ampl (variable) - amplitude of peak;
  100. base (fixed) = 1 Hz - steady-state firing rate after fast exponential depression;
  101. decay (variable) - time-constant of the fast exponential depression;
  102. delay (variable) - delay of the FF input to the cell relative to the stimulus start;
  103. b (fixed) = 0 Hz - baseline activity;
  104. decay_ff (varaible) - time-constant of the FF input exponential decay after the end of stimulation;
  105. s_start (varaible) - delay after stimulus when linear modulation starts;
  106. k (varaible) - slope of the slow linear modulation;
  108. """
  110. if (t < stim + delay):
  111. h = b
  112. elif (t > stim + inter):
  113. h = (b + base + ampl*np.exp(-(inter-delay)*decay)+ (inter - delay - s_start)*k)*np.exp(-(t - stim - inter)*decay_ff)
  114. elif ((t >= stim + delay) and (t < stim + delay + s_start)):
  115. h = b + base + ampl*np.exp(-(t - stim - delay)*decay)
  116. else:
  117. h = b + base + ampl*np.exp(-(t - stim - delay)*decay)+ ((t - stim - delay - s_start)*k)
  119. return h
  121. # %%
  122. def model_step(t,
  123. y,
  124. w_0, w_1, w_2, w_3, w_4, w_5, w_6, w_7, w_8, w_9, w_10, w_11, w_12, w_13, w_14, w_15, w_16, w_17, w_18,
  125. w_19, w_20, w_21, w_22, w_23, tau_0, tau_1, tau_2, tau_3,
  126. threshold,
  127. power,
  128. q,
  129. i_0, i_1, i_2, i_3, i_4,
  130. r_1, decay, delay_1, delay_2, delay_3, ampl, base, decay_s, ampl_1, base_sigm, decay_f, decay_ff, s_start, k
  131. ):
  132. """
  133. Model basis:
  134. ------------------------------------
  135. Systems of first order differential equations that captures activity of populations (df_xdt).
  136. Returns an array of values of the derrivatives at a certain timepoint, calculated from values of functions in previous timepoint.
  137. ------------------------------------
  138. Parameters:
  140. t (running) - time in seconds;
  141. y (running) - values of activities on the previous timepoint;
  142. w_x (variable) - synaptic weights;
  143. tau_x (fixed) - neurons time-constants;
  144. threshold (fixed) = 0 Hz - minimum rectification value;
  145. power (fixed) = 2 - value of the power in the input-output function;
  146. q (fixed) = 1 - normalization coeficient in the input-output function;
  147. i_x (variable) - baseline activity of neurons;
  148. ...
  150. """
  153. f_e, f_p, f_s, f_v_pos, f_v_neg = y
  155. ## Model equations
  157. ff_e = (min(max((i_0 + w_0 * f_e + w_1 * expon(t, ampl = ampl, base = base, decay = decay, delay = delay_1, decay_ff = decay_ff, s_start = s_start, k = k) + w_2 * Sigm(t, ampl = ampl_1, rate = r_1, delay = delay_2, decay_s = decay_s, base = base_sigm) + w_16 * Step(t, k = 0, delay = delay_3, decay_f = decay_f) - w_3 * f_p - w_4 * f_s), threshold), 25))
  158. ff_p = (min(max((i_1 + w_17 * Step(t, k = 0, delay = delay_3, decay_f = decay_f) + w_5 * f_e + w_6 * expon(t, ampl = ampl, base = base, decay = decay, delay = delay_1, decay_ff = decay_ff, s_start = s_start, k = k) + w_7 * Sigm(t, ampl = ampl_1, rate = r_1, delay = delay_2, decay_s = decay_s, base = base_sigm) - w_8 * f_p - w_9 * f_s), threshold), 25))
  159. ff_s = (min(max((i_2 + w_10 * f_e + w_11 * Step(t, k = 0, delay = delay_3, decay_f = decay_f) - w_12 * f_v_pos - w_23 * f_v_neg), threshold), 25))
  160. ff_v_pos = (min(max((i_3 + w_18 * Step(t, k = 0, delay = delay_3, decay_f = decay_f) + w_13 * f_e - w_14 * f_s + w_15 * Sigm(t, ampl = ampl_1, rate = r_1, delay = delay_2, decay_s = decay_s, base = base_sigm)), threshold), 25))
  161. ff_v_neg = (min(max((i_4 + w_19 * Step(t, k = 0, delay = delay_3, decay_f = decay_f) + w_20 * f_e - w_21 * f_s + w_22 * Sigm(t, ampl = ampl_1, rate = r_1, delay = delay_2, decay_s = decay_s, base = base_sigm)), threshold), 25))
  163. df_edt = ((q * ff_e ** power) - f_e) / tau_0
  164. df_pdt = ((q * ff_p ** power) - f_p) / tau_1
  165. df_sdt = ((q * ff_s ** power) - f_s) / tau_2
  166. df_v_posdt = ((q * ff_v_pos ** power) - f_v_pos) / tau_3
  167. df_v_negdt = ((q * ff_v_neg ** power) - f_v_neg) / tau_3
  169. dydt = [df_edt, df_pdt, df_sdt, df_v_posdt, df_v_negdt]
  171. return dydt
  173. # %%
  174. names = ['PC_PC', # w_0
  175. 'FF_PC', # w_1
  176. 'SS_PC', # w_2
  177. 'PV_PC', # w_3
  178. 'SST_PC', # w_4
  179. 'PC_PV', # w_5
  180. 'FF_PV', # w_6
  181. 'SS_PV', # w_7
  182. 'PV_PV', # w_8
  183. 'SST_PV', # w_9
  184. 'PC_SST', # w_10
  185. 'FB_SST', # w_11
  186. 'VIP+_SST', # w_12
  187. 'PC_VIP+', # w_13
  188. 'SST_VIP+', # w_14
  189. 'SS_VIP+', # w_15
  190. 'FB_PC', # w_16
  191. 'FB_PV', # w_17
  192. 'FB_VIP+', # w_18
  193. 'FB_VIP-', # w_19
  194. 'PC_VIP-', # w_20
  195. 'SST_VIP-', # w_21
  196. 'SS_VIP-', # w_22
  197. 'VIP-_SST']# w_23
  201. # %%
  202. def exp_time(start, step, count, endpoint=False):
  203. """
  204. Experimental timepoints calculation:
  205. ------------------------------------
  206. Returns an array of values of the experimental timepoints.
  207. ------------------------------------
  208. Parameters:
  210. start - starting point;
  211. step - value of time step of experimental recordings;
  212. count - number of points;
  214. """
  215. stop = start+(step*count)
  216. return np.linspace(start, stop, count, endpoint=endpoint)
  218. # %%
  219. def odesol_step(tt, init, params):
  220. """
  221. Solves differential equation system defined in model_step() function.
  222. """
  223. y_init = init
  224. w_0 = params['w_0'].value
  225. w_1 = params['w_1'].value
  226. w_2 = params['w_2'].value
  227. w_3 = params['w_3'].value
  228. w_4 = params['w_4'].value
  229. w_5 = params['w_5'].value
  230. w_6 = params['w_6'].value
  231. w_7 = params['w_7'].value
  232. w_8 = params['w_8'].value
  233. w_9 = params['w_9'].value
  234. w_10 = params['w_10'].value
  235. w_11 = params['w_11'].value
  236. w_12 = params['w_12'].value
  237. w_13 = params['w_13'].value
  238. w_14 = params['w_14'].value
  239. w_15 = params['w_15'].value
  240. w_16 = params['w_16'].value
  241. w_17 = params['w_17'].value
  242. w_18 = params['w_18'].value
  243. w_19 = params['w_19'].value
  244. w_20 = params['w_20'].value
  245. w_21 = params['w_21'].value
  246. w_22 = params['w_22'].value
  247. w_23 = params['w_23'].value
  249. tau_0 = params['tau_0'].value
  250. tau_1 = params['tau_1'].value
  251. tau_2 = params['tau_2'].value
  252. tau_3 = params['tau_3'].value
  253. threshold = params['threshold'].value
  254. power = params['power'].value
  255. q = params['q'].value
  256. i_0 = params['i_0'].value
  257. i_1 = params['i_1'].value
  258. i_2 = params['i_2'].value
  259. i_3 = params['i_3'].value
  260. i_4 = params['i_4'].value
  261. ampl_1 = params['ampl_1'].value
  262. r_1 = params['r_1'].value
  263. delay_1 = params['delay_1'].value
  264. delay_2 = params['delay_2'].value
  265. delay_3 = params['delay_3'].value
  266. decay = params['decay'].value
  267. decay_s = params['decay_s'].value
  268. decay_f = params['decay_f'].value
  269. decay_ff = params['decay_ff'].value
  270. ampl = params['ampl'].value
  271. base = params['base'].value
  272. base_sigm = params['base_sigm'].value
  273. s_start = params['s_start'].value
  274. k = params['k'].value
  276. sol = solve_ivp(lambda t, y: model_step(t, y,
  277. w_0, w_1, w_2, w_3, w_4, w_5, w_6, w_7, w_8, w_9, w_10, w_11, w_12, w_13, w_14, w_15, w_16, w_17, w_18,
  278. w_19, w_20, w_21, w_22, w_23, tau_0, tau_1, tau_2, tau_3,
  279. threshold,
  280. power,
  281. q,
  282. i_0, i_1, i_2, i_3, i_4,
  283. r_1, decay, delay_1, delay_2, delay_3, ampl, base, decay_s, ampl_1, base_sigm, decay_f, decay_ff, s_start, k
  284. ),
  285. [tt[0], tt[-1]],
  286. y_init,
  287. method='RK45',
  288. t_eval=tt,
  289. )
  291. return sol
  293. # %%
  294. def simulate_step(tt, init, params):
  295. """
  296. Model simulation:
  297. ------------------------------------
  298. Returns a Pandas DataFrame of timesteps and populational activity traces.
  299. ------------------------------------
  300. Parameters:
  302. tt (running) - current timepoint;
  303. init (fixed) - intial conditions for differential equations;
  304. params - lmfit.Parameters() object with all parameters from the differential equations systems
  305. """
  307. sol = odesol_step(tt, init, params)
  309. dd = np.vstack((sol.t, sol.y[0], sol.y[1], sol.y[2], sol.y[3], sol.y[4])).T
  310. sim = pd.DataFrame(dd, columns=['t', 'f_e', 'f_pv', 'f_sst', 'f_vip_pos', 'f_vip_neg'])
  312. return sim
  314. # %%
  315. def RMSE_full(params, init, data_pc, data_pv, data_sst, data_vip_pos, data_vip_neg):
  316. """
  317. Root:Mean Square Error (RMSE) calculation:
  318. ------------------------------------
  319. Returns RMSE value for full fit of 4 averages of populations: PC, PV, SST, VIP.
  320. ------------------------------------
  321. Parameters:
  323. params - Parameters() object with all parameters from the differential equations systems
  324. init (fixed) - intial conditions for differential equations;
  325. data_pc - PC experimental data trace;
  326. data_pv - PV experimental data trace;
  327. data_sst - SST experimental data trace;
  328. data_vip_pos - VIP Positive experimental data trace;
  329. data_vip_neg - VIP Negative experimental data trace;
  331. """
  332. model = simulate_step(t_exp, init, params)
  333. sum = 0
  334. for i in range(len(data_pc)):
  335. sum += (model['f_e'].values[i] - data_pc[i])**2 + (model['f_pv'].values[i] - data_pv[i])**2 + (model['f_sst'].values[i] - data_sst[i])**2 + (model['f_vip_pos'].values[i] - data_vip_pos[i])**2 + (model['f_vip_neg'].values[i] - data_vip_neg[i])**2
  337. sum_norm = np.sqrt((sum)/(len(data_pc)*5))
  339. return sum_norm
  341. # %% [markdown]
  342. # # Experimental data
  344. # %%
  345. #timepoint
  346. dt = 0.164745
  348. # PC_average (whole population)
  349. pc_all = np.loadtxt(r'Experimental_Data\QA_Nor_PC_All_AvgSess1.txt')
  351. # Experimental time calculation
  352. t_exp_1 = exp_time(start=0.0, step=dt, count=len(pc_all))
  354. # PV_average (whole population)
  355. pv = np.loadtxt(r'Experimental_Data\QA_Nor_PV_AvgSess1.txt')
  357. # SST_average (whole population)
  358. sst = np.loadtxt(r'Experimental_Data\QA_Nor_SST_AvgSess1.txt')
  360. # VIP_Positive (whole population)
  361. vip_pos = np.loadtxt(r'Experimental_Data\QA_Nor_VIP_AvgSess1_Pos.txt')
  363. # VIP_Negative (whole population)
  364. vip_neg = np.loadtxt(r'Experimental_Data\QA_Nor_VIP_AvgSess1_Neg.txt')
  366. # %%
  367. # Inicialization of new arrays to cut traces for suitable range
  369. t_exp_1_new = []
  370. pv_new = []
  371. sst_new = []
  372. vip_pos_new = []
  373. vip_neg_new = []
  374. pc_all_new = []
  377. # Cutting the traces
  378. for ind, time in np.ndenumerate(t_exp_1):
  379. if (time >= 5) and (time <= 25):
  380. t_exp_1_new.append(time - 5)
  381. pv_new.append(pv[ind])
  382. sst_new.append(sst[ind])
  383. vip_pos_new.append(vip_pos[ind])
  384. vip_neg_new.append(vip_neg[ind])
  385. pc_all_new.append(pc_all[ind])
  387. # Adjusting for a missing initial point after deconvolution
  388. pv_new.insert(0, pv_new[0])
  389. sst_new.insert(0, sst_new[0])
  390. vip_pos_new.insert(0, vip_pos_new[0])
  391. vip_neg_new.insert(0, vip_neg_new[0])
  392. pc_all_new.insert(0, pc_all_new[0])
  394. t_exp_1_new.append(t_exp_1_new[-1] + dt)
  396. # %% [markdown]
  397. # # Model parameters
  399. # %%
  400. # ===============================================================
  401. # Setting initial and working parameters
  402. # ===============================================================
  404. #Fixed parameters
  406. t = np.float32(np.array(t_exp_1_new))
  407. t_exp = t
  408. threshold = np.float32(0)
  409. power = np.float32(2.0)
  410. q = np.float32(1.0)
  411. ampl_1 = 1.0
  412. r_1 = np.float32(1.713939)
  413. delay_1 = np.float32(0.1650418)
  414. delay_2 = np.float32(1.73305)
  415. delay_3 = np.float32(0.3695274)
  416. decay = np.float32(3.324089)
  417. decay_s = np.float32(0.2685497)
  418. decay_f = np.float32(0.8342502)
  419. decay_ff = np.float32(2.152722)
  420. ampl = np.float32(1.0)
  421. base = np.float32(1.0)
  422. base_sigm = np.float32(0)
  423. s_start = np.float32(0.1835884)
  424. init = np.float32(np.array([0.055, 0.011, 0.017, 0.208, 1.000]))
  425. tau = np.float32(np.array([0.015, 0.0075, 0.019, 0.019]))
  427. #Variable parameters
  429. k = np.float32(0.027309763)
  430. i = np.float32(np.array([0.284447361, 0.011142876, 0.116241266, 0.50467562, 0.942411609]))
  431. w = np.float32(np.array([
  432. 0.080753061, # w_0 PC -> PC
  433. 1.836526807, # w_1 FF -> PC
  434. 0.784303879, # w_2 SS -> PC
  435. 2.063944459, # w_3 PV -> PC
  436. 0.748021346, # w_4 SST -> PC
  437. 0.388020396, # w_5 PC -> PV
  438. 1.142616121, # w_6 FF -> PV
  439. 1.791791214, # w_7 SS -> PV
  440. 1.595357309, # w_8 PV -> PV
  441. 0.334000211, # w_9 SST -> PV
  442. 0.904594776, # w_10 PC -> SST
  443. 0.493489103, # w_11 FB -> SST
  444. 0.159572032, # w_12 VIP+ -> SST
  445. 0.254720976, # w_13 PC -> VIP+
  446. 0.167011082, # w_14 SST -> VIP+
  447. 0.692450185, # w_15 SS -> VIP+
  448. 1.35949409, # w_16 FB -> PC
  449. 0.304908259, # w_17 FB -> PV
  450. 0.232536359, # w_18 FB -> VIP+
  451. 0.146459393, # w_19 FB -> VIP-
  452. 0.3620719, # w_20 PC -> VIP-
  453. 0.727962823, # w_21 SST -> VIP-
  454. 0.174327124, # w_22 SS -> VIP-
  455. 0.000760818]))# w_23 VIP- -> SST
  458. data_pv = np.float32(np.array(pv_new))
  459. data_sst = np.float32(np.array(sst_new))
  460. data_vip_pos = np.float32(np.array(vip_pos_new))
  461. data_vip_neg = np.float32(np.array(vip_neg_new))
  462. data_pc_all = np.float32(np.array(pc_all_new))
  464. count = 0
  466. # %%
  467. # ============================
  468. # Parameter definition
  469. # ============================
  471. params = Parameters()
  472. params.add('w_0', value = w[0])
  473. params.add('w_1', value = w[1])
  474. params.add('w_2', value = w[2])
  475. params.add('w_3', value = w[3])
  476. params.add('w_4', value = w[4])
  477. params.add('w_5', value = w[5])
  478. params.add('w_6', value = w[6])
  479. params.add('w_7', value = w[7])
  480. params.add('w_8', value = w[8])
  481. params.add('w_9', value = w[9])
  482. params.add('w_10', value = w[10])
  483. params.add('w_11', value = w[11])
  484. params.add('w_12', value = w[12])
  485. params.add('w_13', value = w[13])
  486. params.add('w_14', value = w[14])
  487. params.add('w_15', value = w[15])
  488. params.add('w_16', value = w[16])
  489. params.add('w_17', value = w[17])
  490. params.add('w_18', value = w[18])
  491. params.add('w_19', value = w[19])
  492. params.add('w_20', value = w[20])
  493. params.add('w_21', value = w[21])
  494. params.add('w_22', value = w[22])
  495. params.add('w_23', value = w[23])
  497. params.add('tau_0', value = tau[0])
  498. params.add('tau_1', value = tau[1])
  499. params.add('tau_2', value = tau[2])
  500. params.add('tau_3', value = tau[3])
  501. params.add('threshold', value = threshold)
  502. params.add('power', value = power)
  503. params.add('q', value = q)
  504. params.add('i_0', value = i[0])
  505. params.add('i_1', value = i[1])
  506. params.add('i_2', value = i[2])
  507. params.add('i_3', value = i[3])
  508. params.add('i_4', value = i[4])
  510. params.add('ampl_1', value = ampl_1)
  511. params.add('r_1', value = r_1)
  512. params.add('delay_1', value = delay_1)
  513. params.add('delay_2', value = delay_2)
  514. params.add('delay_3', value = delay_3)
  515. params.add('decay', value = decay)
  516. params.add('decay_s', value = decay_s)
  517. params.add('decay_f', value = decay_f)
  518. params.add('decay_ff', value = decay_ff)
  519. params.add('ampl', value = ampl)
  520. params.add('base', value = base)
  521. params.add('base_sigm', value = base_sigm)
  522. params.add('s_start', value = s_start)
  523. params.add('k', value = k)
  525. # %% [markdown]
  526. # ## Model Simulation
  528. # %% [markdown]
  529. # ### Connection weights (wji)
  531. # %%
  532. vals = []
  533. for _, param in params.items():
  534. vals.append(param.value)
  536. plt.figure(figsize=(15,7.5))
  537. plt.ylabel(r"$w_{ji}$", fontsize=20)
  538. plt.bar(range(len(vals[:24])),vals[:24], color='blue')
  539. plt.xticks(range(len(vals[:24])), names[:24], rotation = 60, fontsize=15)
  540. plt.bar_label(plt.bar(range(len(vals[:24])),vals[:24]), fmt = '{:.2f}', fontsize=15)
  541. plt.yticks(np.arange(0,2.5,0.5), fontsize=15)
  542. plt.show()
  544. # %%
  545. # Simulate the model with parameters defined in params
  546. model = simulate_step(t_exp, init, params)
  548. # %% [markdown]
  549. # ### Fitting to experimental traces
  551. # %%
  552. # ============================
  553. # PC
  554. # ============================
  555. plt.figure(figsize=(4,4))
  556. plt.plot(t_exp, data_pc_all, color='#000000', marker='o', label='PC data')
  557. plt.plot(t_exp, model['f_e'].values, color='#ffaa00', label='fit', linewidth=3)
  558. plt.title('PC')
  559. plt.legend(loc='best')
  560. plt.ylabel("Spike Rate (Hz)", fontsize=14)
  561. plt.xlabel('t', fontsize=14)
  562. plt.xticks(np.arange(0,21,5))
  563. plt.yticks(np.arange(0,1.7,0.5))
  564. plt.show()
  566. # %%
  567. # ============================
  568. # VIP Positive
  569. # ============================
  570. plt.figure(figsize=(4,4))
  571. plt.plot(t_exp, data_vip_pos, color='#02ab81', marker='o', label='VIP Pos data')
  572. plt.plot(t_exp, model['f_vip_pos'].values, color='#ffaa00', label='fit', linewidth=3)
  573. plt.title('VIP Positive')
  574. plt.legend(loc='best')
  575. plt.ylabel("Spike Rate (Hz)", fontsize=14)
  576. plt.xlabel('t', fontsize=14)
  577. plt.xticks(np.arange(0,21,5))
  578. plt.yticks(np.arange(0,2.2,0.5))
  579. plt.show()
  581. # %%
  582. # ============================
  583. # VIP Negative
  584. # ============================
  585. plt.figure(figsize=(4,4))
  586. plt.plot(t_exp, data_vip_neg, color='#02ab81', marker='o', label='VIP Neg data')
  587. plt.plot(t_exp, model['f_vip_neg'].values, color='#ffaa00', label='fit', linewidth=3)
  588. plt.title('VIP Negative')
  589. plt.legend(loc='best')
  590. plt.ylabel("Spike Rate (Hz)", fontsize=14)
  591. plt.xlabel('t', fontsize=14)
  592. plt.xticks(np.arange(0,21,5))
  593. plt.yticks(np.arange(0,2.2,0.5))
  594. plt.show()
  596. # %%
  597. # ============================
  598. # PV
  599. # ============================
  600. plt.figure(figsize=(4,4))
  601. plt.plot(t_exp, data_pv, color='#ae00ff', marker='o', label='PV data')
  602. plt.plot(t_exp, model['f_pv'].values, color='#ffaa00', label='fit', linewidth=3)
  603. plt.title('PV')
  604. plt.legend(loc='best')
  605. plt.ylabel("Spike Rate (Hz)", fontsize=14)
  606. plt.xlabel('t', fontsize=14)
  607. plt.xticks(np.arange(0,21,5))
  608. plt.yticks(np.arange(0,1.7,0.4))
  609. plt.show()
  611. # %%
  612. # ============================
  613. # SST
  614. # ============================
  615. plt.figure(figsize=(4,4))
  616. plt.plot(t_exp, data_sst, color='#d80000', marker='o', label='SST data')
  617. plt.plot(t_exp, model['f_sst'].values, color='#ffaa00', label='fit', linewidth=3)
  618. plt.title('SST')
  619. plt.legend(loc='best')
  620. plt.ylabel("Spike Rate (Hz)", fontsize=14)
  621. plt.xlabel('t', fontsize=14)
  622. plt.xticks(np.arange(0,21,5))
  623. plt.yticks(np.arange(0,2.8,0.5))
  624. plt.show()
  626. # %% [markdown]
  627. # ### Root mean square error calculation (RMSE)
  629. # %%
  630. a_tot = RMSE_full(params, init, data_pc_all, data_pv, data_sst, data_vip_pos, data_vip_neg)
  631. print(a_tot)
  633. # %% [markdown]
  634. # ### Saving Traces
  636. # %%
  637. # Write csv file with output. f_e, f_p, f_s, f_v, f_e_o, f_p_o, f_s_o, f_v_o = y
  638. model[['t', 'f_e']].to_csv('Trace_PC_HabSess1.csv', index=False)
  639. model[['t', 'f_pv']].to_csv('Trace_PV_HabSess1.csv', index=False)
  640. model[['t', 'f_sst']].to_csv('Trace_SST_HabSess1.csv', index=False)
  641. model[['t', 'f_vip_pos']].to_csv('Trace_VIPPos_HabSess1.csv', index=False)
  642. model[['t', 'f_vip_neg']].to_csv('Trace_VIPNeg_HabSess1.csv', index=False)

Model_V1_Hab_Sess1-checkpoint.ipynb at commit 5d14a6b, under Apache-2.0 · at the source

Overview

Authors: A.J. Hinojosa1, S.E. Dominiak1, Ye. Kosiachkin1, L. Lagnado1
ORCID iDs: A.J. Hinojosa, S.E. Dominiak, Ye. Kosiachkin
  1. Sussex Neuroscience, School of Life Sciences, University of Sussex, Brighton, BN1 9QG, United Kingdom
Institutions: University of Sussex (United Kingdom)
Journal: bioRxiv (preprint)
Dates: published online 25 March 2026
Type: Preprint
License: CC BY
Identifiers: DOI 10.64898/2026.03.24.713929 · OpenAlex W7140810994
Open access: green, a free copy (OpenAlex)
Status: code verified (how the code is found and verified)
Methods: Statistics, Connectivity, fMRI & imaging, Single-unit activity, calcium imaging, Physiology & signal measures
Topic: Neural dynamics and brain function (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: Biotechnology and Biological Sciences Research Council (BB/X009386/1)
Citations: not cited yet (Europe PMC); 59 references in the paper

Abstract

Sensory cortices filter repeated inputs through rapid adaptation over seconds and experience-driven learning over days. Although these forms of plasticity occur simultaneously, it is not known how they interact within cortical circuits. We combined two-photon calcium imaging, data-driven circuit modelling and optogenetics to investigate how short-term adaptation in layer 2/3 of mouse V1 is shaped by stimulus familiarity and reward association. Habituation reduced the fraction of pyramidal cells responsive to a visual stimulus, whereas reward association maintained overall responsivity. In contrast, both forms of learning shifted pyramidal cell adaptation from depression toward sensitization, but through distinct circuit mechanisms. Habituation reduced disinhibition through the VIP→SST→PC pathway by weakening feedback activation of VIPs and VIP→SST connections. Reward association counteracted this effect by increasing disinhibition through the SST→PV→PC pathway, strengthening SST→PV connections while reducing SST→PC inputs. Despite engaging distinct disinhibitory circuits and producing divergent effects on pyramidal cell responsivity, both forms of learning converged on a reduced PV:SST input ratio to pyramidal cells, thereby biasing V1 toward sensitizing adaptation. These results identify changes in cortical circuits that link the plasticity of fast adaptation to simple forms of learning.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above.

lagnadoLab/CortexModel-Learning

License: Apache-2.0
State: the link answers, verified on 30 September 2026
Evidence: files inventoried
Commit: 5d14a6b64a45586ce2fcdc96376f4d4dfea102eb, 24 March 2026
Languages: Python (24), Jupyter (8), Shell (8)
Size: 94 files, 40 scripts
Software Heritage: not archived
Found in: the text, “Model”
Holds: README, license file, environment (requirements.txt, Fitting/requirements.txt), 4 notebooks
Not found: CITATION.cff, tests, continuous integration, documentation
Tools: pandas (32 files), NumPy (28 files), SciPy (24 files), LMFIT (16 files), Matplotlib (16 files), scikit-learn (12 files), seaborn (8 files)
Availability: 1 check, the latest on 30 September 2026: the link answers
  • 30 September 2026: the link answers
42 files
At the source: github.com/lagnadoLab/CortexModel-Learning

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 40 scripts, each with its path and the digest of its content;
  • no match between paragraphs and code yet;
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 30 September 2026: the first record

Recorded: type, journal, dates, 4 authors, 1 funder, 56 references.

Cite

This paper

Hinojosa, A., Dominiak, S., Kosiachkin, Y., & Lagnado, L. (2026). Distinct Disinhibitory Circuits Link Short-Term Adaptation to Familiarity and Reward Learning in Visual Cortex. bioRxiv (preprint). https://doi.org/10.64898/2026.03.24.713929

BibTeX

@article{hinojosa2026distinct,
author = {Hinojosa, A.J. and Dominiak, S.E. and Kosiachkin, Ye. and Lagnado, L.},
title = {{Distinct Disinhibitory Circuits Link Short-Term Adaptation to Familiarity and Reward Learning in Visual Cortex}},
journal = {bioRxiv (preprint)},
year = {2026},
month = mar,
publisher = {bioRxiv},
issn = {2692-8205},
doi = {10.64898/2026.03.24.713929},
url = {https://doi.org/10.64898/2026.03.24.713929}
}

RIS

TY - JOUR
AU - Hinojosa, A.J.
AU - Dominiak, S.E.
AU - Kosiachkin, Ye.
AU - Lagnado, L.
TI - Distinct Disinhibitory Circuits Link Short-Term Adaptation to Familiarity and Reward Learning in Visual Cortex
T2 - bioRxiv (preprint)
J2 - bioRxiv
PY - 2026
DA - 2026/03/25
SN - 2692-8205
PB - bioRxiv
DO - 10.64898/2026.03.24.713929
UR - https://doi.org/10.64898/2026.03.24.713929
ER -

CSL-JSON

{
"id": "10.64898/2026.03.24.713929",
"type": "article",
"title": "Distinct Disinhibitory Circuits Link Short-Term Adaptation to Familiarity and Reward Learning in Visual Cortex",
"container-title": "bioRxiv (preprint)",
"author": [
{
"family": "Hinojosa",
"given": "A.J."
},
{
"family": "Dominiak",
"given": "S.E."
},
{
"family": "Kosiachkin",
"given": "Ye."
},
{
"family": "Lagnado",
"given": "L."
}
],
"container-title-short": "bioRxiv",
"DOI": "10.64898/2026.03.24.713929",
"ISSN": "2692-8205",
"publisher": "bioRxiv",
"URL": "https://doi.org/10.64898/2026.03.24.713929",
"issued": {
"date-parts": [
[
2026,
3,
25
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1371/journal.pbio.3003518 [code]
Context-dependent activation of V1 parvalbumin interneurons enhances visual discrimination.
Journal: PLoS biology
In common: pandas, SciPy, Matplotlib, 1 other tool, 9 references
[2] doi:10.1126/sciadv.aed6417 [code]
Intrinsic timing, not temporal prediction, underlies ramping dynamics in visual and parietal cortex during passive behavior.
Journal: Science advances
In common: seaborn, scikit-learn, pandas, 3 other tools, 6 references
[3] doi:10.1523/jneurosci.0987-25.2026 [code]
Cell-Type-Specific Synaptic Scaling Mechanisms Differentially Contribute to Associative Learning.
Journal: The Journal of neuroscience : the official journal of the Society for Neuroscience
In common: seaborn, pandas, SciPy, 2 other tools, 7 references
[4] doi:10.1101/2025.05.14.654005 [code]
Cell-type-specific synaptic scaling mechanisms differentially contribute to associative learning
Journal: bioRxiv (preprint)
In common: seaborn, pandas, SciPy, 2 other tools, 7 references
[5] doi:10.1093/cercor/bhag047 [code]
Association learning drives synaptic plasticity at feedforward synapses in somatosensory cortex.
Journal: Cerebral cortex (New York, N.Y. : 1991)
In common: scikit-learn, pandas, SciPy, 2 other tools, 6 references
[6] doi:10.1073/pnas.2312898121 [code]
A disinhibitory circuit mechanism explains a general principle of peak performance during mid-level arousal.
Journal: Proceedings of the National Academy of Sciences of the United States of America
In common: LMFIT, seaborn, scikit-learn, 4 other tools, 3 references
[7] doi:10.1016/j.celrep.2023.113088
VIP interneurons regulate cortical size tuning and visual perception.
Journal: Cell reports
In common: 8 references
[8] doi:10.1016/j.isci.2025.114132 [code]
FMRP restoration in parvalbumin interneurons: A circuit-specific improvement of visual learning in fragile X syndrome.
Journal: iScience
In common: seaborn, scikit-learn, pandas, 3 other tools, 4 references
[9] doi:10.1073/pnas.2116895119 [code]
A circuit mechanism for independent modulation of excitatory and inhibitory firing rates after sensory deprivation.
Journal: Proceedings of the National Academy of Sciences of the United States of America
In common: Matplotlib, NumPy, 7 references
[10] doi:10.1038/s41467-025-65665-4 [code]
Disrupted theta synchronization and synaptic connectivity in the visual cortex of Fmr1 KO mice.
Journal: Nature communications
In common: seaborn, scikit-learn, pandas, 3 other tools, 4 references

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Add the badge to the code's README

OSCR: the badge of this paper

If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.

Markdown
<!-- oscr-badge:start -->
[![OSCR](https://openscicode.org/badge/doi/10.64898/2026.03.24.713929.svg)](https://openscicode.org/badge/doi/10.64898/2026.03.24.713929)
<!-- oscr-badge:end -->
reStructuredText
.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.64898/2026.03.24.713929.svg
   :target: https://openscicode.org/badge/doi/10.64898/2026.03.24.713929
   :alt: OSCR

.. oscr-badge:end
HTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.64898/2026.03.24.713929"><img src="https://openscicode.org/badge/doi/10.64898/2026.03.24.713929.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->

On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".

If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Report a mismatch between the code and the paper · Report a code error (signed in)

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Report a reproduction failure (signed in)

Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.