OSCR

Subspace reverse-correlation estimation of receptive fields during free viewing.

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, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

MATLAB · 712 lines · 16 KB · CC-BY-4.0

  1. %% This is the code for eye movement correction in the paper "Subspace reverse-correlation estimation of receptive fields during free viewing"
  2. %% Loading the data
  3. clc;
  4. clear;
  5. experiment='np01_003_000'; % Experiment name
  6. data=load(strcat('./data/',experiment,'.db'),'-mat'); % loading the log data
  8. % dimensions of the visual field (the monitor). Here we downsampled our monitor size by a factor of 8.
  9. RF_size=[480,135];
  11. %% Computing the frequency domain receptive fields (fRFs)
  13. [fRFs,norms]=krnls(data.log); % Computing the frequency domain receptive fields (fRFs)
  14. fRFs=mean(fRFs,3);
  16. %% Selecting cells to pass to the optimization algorithm
  18. % FIRST METHOD
  19. % If the RFs are strongly distorted by eye movements such that they are not detectable we may select the cells based on the fRFs and their frequency.
  20. % [poris,freqs,frfm]=freq_extract(fRFs); % Extracting preferred oreintations (poris), preferred spatial frequencies (freqs)
  22. % index=find(frfm>0.01 & (freqs<16)); % finding the cells with large enough mean response and preferred frequency below a threshold
  23. % select=index; % The select indices will be used for optimization.
  24. % mask=[]; % empty mask allows optimization to be applied to all the visual field domain.
  26. % SECOND METHOD
  27. % If the RFs are visible we can select cells based on the Riesz envlope sharpness of the RFs.
  28. % If needed a mask can be used to limit the optimizer to particular areas of the receptive fields.
  30. % The following code gives us mask and selected neurons in mouse data.
  31. m=0.5;
  32. [mask,index,env,Narea,Nmax]=getmasks(data.log,m,RF_size); % Riesz envelopes are returned as env. Masks are seleced by thresholding the envelopes above 50 percent of their max value.
  33. select=index(1:40);
  34. disp('selected=')
  35. disp(select)
  36. mask=[]; % empty mask allows optimization to be applied to all the visual field domain.
  38. % SECOND METHOD
  39. % You can select the neurons and masks by hand.
  40. % select must be a one dimensional array with indices of selected neurons
  41. % mask must be a 3D array of zeros and ones with first index for neurons and other two
  42. % equivalent to dimensions of the screen. The optimization will be applied to mask==1.
  43. %% running the optimization algorithm
  44. A0=[4.5,0 ; 0,4.5]; % initial guess of the A matrix.
  45. [A,fval,exitflag,output,grad]=EYE_CORRECT(data.log,select,mask,A0,RF_size);
  47. %% Evaluating all RFs before and after correction
  48. tic
  49. rfsb=RECEPTIVE_FIELDS(data.log,[],[],0,RF_size); % before correction
  50. rfsa=RECEPTIVE_FIELDS(data.log,[],A,0,RF_size); % after correction
  51. toc
  54. % rearranging the data
  55. rfsa= permute(rfsa,[1 2 4 3]);
  56. rfsb= permute(rfsb,[1 2 4 3]);
  60. %% visualizing the RFs
  61. ftsz=18;
  62. N=length(select);
  63. % disp(select);
  65. for i=1:N
  67. ii=select(i);
  68. % rfb=squeeze(mean(squeeze(rfsb(:,ii,:,:)),1));
  69. % rfa=squeeze(mean(squeeze(rfsa(:,ii,:,:)),1));
  71. rfb=squeeze(squeeze(rfsb(ii,:,:)))';
  72. rfa=squeeze(squeeze(rfsa(ii,:,:)))';
  73. bmax=max(rfb(:));
  74. amax=max(rfa(:));
  75. cmax=max(bmax,amax);
  78. bmin=min(rfb(:));
  79. amin=min(rfa(:));
  80. cmin=min(bmin,amin);
  81. figure(1)
  82. ax1=subplot(2,2,2);
  83. imagesc(rfb);
  84. axis equal
  85. axis tight
  86. % axis off
  87. set(gca,'xticklabel',[])
  88. set(gca,'yticklabel',[])
  89. set(gca,'XTick',[])
  90. set(gca,'YTick',[])
  91. set(gca,'YDir','normal')
  93. colormap(ax1,parula)
  94. clim([cmin cmax])
  95. title('Naive','fontsize',ftsz)
  97. colorbar
  98. ax2=subplot(2,2,4);
  99. imagesc(rfa);
  100. axis equal
  101. axis tight
  102. % axis off
  103. set(gca,'xticklabel',[])
  104. set(gca,'yticklabel',[])
  105. set(gca,'XTick',[])
  106. set(gca,'YTick',[])
  107. set(gca,'YDir','normal')
  108. colormap(ax1,parula)
  109. clim([cmin cmax])
  110. title('Corrected','fontsize',ftsz)
  111. colorbar
  112. subplot(1,2,1)
  113. imagesc(squeeze(fRFs(:,:,ii)))
  114. axis equal
  115. axis tight
  116. xlabel('$k$','interpreter','latex','fontsize',ftsz)
  117. ylabel('$l$','interpreter','latex','fontsize',ftsz)
  118. title('fRF','fontsize',ftsz)
  119. pause;
  122. end
  128. %%
  131. function [A,fval,exitflag,output,grad,mask]=EYE_CORRECT(DATA,select,mask,A0,RF_size)
  133. % The cost function is based on the norm^2 of the receptive
  134. % fields.
  137. % Inputs are:
  139. % DATA=log
  141. % "select" is the neurons selected to do the optimization based on their
  142. % receptive fields. If select = [] all the neurons will be selected
  144. % mask allows focusing on the receptive fields is a binary array of zeros
  145. % and ones. zero outside of the RF domain and one in the RF domain of each
  146. % neuron. If mask=[] the code uses all the visual field to calibrate eye
  147. % movements
  149. % A0 is the initial guess for 2by2 matrix A. if A=[] then A0=[4,0 ; 0,4]
  150. % will be used
  153. % Outputs are:
  154. % A, fval , exitflag , output , grad at the optimum point.
  156. % Putting data to the format to pass to optimization algorithm
  157. disp('Putting data to the format to pass to optimization algorithm');
  158. ncells = size(DATA.signal{1},2);
  159. nstim = size(DATA,1);
  160. x_E=DATA.xpos-mean(DATA.xpos,'omitnan');
  161. y_E=DATA.ypos-mean(DATA.ypos,'omitnan');
  163. if(sum(isnan(x_E))+sum(isnan(y_E))>0)
  164. disp('--------------------------------')
  165. warning('There are trials with NAN values for eye positions that are set to zero')
  166. disp('--------------------------------')
  167. end
  169. idxnanx=isnan(x_E);
  170. idxnany=isnan(y_E);
  171. x_E(idxnanx)=0;
  172. y_E(idxnany)=0;
  176. S_tr=zeros(nstim,3);
  178. S_tr(:,1)=DATA.kx;
  179. S_tr(:,2)=DATA.ky;
  180. S_tr(:,3)=DATA.sign;
  182. R=cat(3,DATA.signal{:});
  183. Lambda_tr=squeeze(mean(R(2:7,:,:),1));
  186. if(isempty( select ))
  187. select=1:ncells;
  188. end
  190. if(isempty( A0 ))
  191. A0=[4,0 ; 0,4];
  192. end
  194. r=Lambda_tr(select,:);
  196. if(isempty(mask))
  197. mask1=[];
  198. else
  199. mask1=mask(select,:,:);
  200. end
  204. % running the minimization algorithm
  205. X0=[ A0(1,1) , A0(1,2) , A0(2,1) , A0(2,2) ];
  206. disp('starting optimization')
  207. tic
  208. opt = optimoptions('fminunc','SpecifyObjectiveGradient',true,'Display','iter','PlotFcns',@optimplotfval,'MaxIter',150,'TolFun',1e-5,'TolX',1e-5,'OutputFcn', @outfun);
  209. [X,fval,exitflag,output,grad] = fminunc( @(X)EYE_CORRECT_COSTSD(X,x_E,y_E,S_tr,r,RF_size,mask1) , X0 , opt );
  210. toc
  212. A=[X(1),X(2);X(3),X(4)];
  214. end
  219. %% Functions
  221. function U=KXLY(mon_size,k,l)
  222. % meshgrid phase of the cas function in the visual field
  223. Xmon=mon_size(1);
  224. Ymon=mon_size(2);
  225. ix=(1:Xmon);
  226. iy=(1:Ymon);
  227. [X,Y]=meshgrid(ix,iy);
  228. U=(2*pi/Xmon)*(k*X+l*Y);
  229. U=U';
  230. end
  234. %% This is a function that decouples the space and time components of the cas
  237. function [Rcos,Rsin,DaRcos,DaRsin,DbRcos,DbRsin,DcRcos,DcRsin,DdRcos,DdRsin]=reorganize(Lambda_tr,S_tr,x_E,y_E,A,RF_size)
  238. a=A(1,1);
  239. b=A(1,2);
  240. c=A(2,1);
  241. d=A(2,2);
  243. [N,~]=size(Lambda_tr);
  244. k=unique(S_tr(:,1));
  245. l=unique(S_tr(:,2));
  246. p=unique(S_tr(:,3));
  249. Nk=length(k);
  250. Nl=length(l);
  251. Np=length(p);
  253. XRF=RF_size(1);
  255. Rsin=zeros(Nk,Nl,Np,N);
  256. Rcos=zeros(Nk,Nl,Np,N);
  257. DaRcos=zeros(Nk,Nl,Np,N);
  258. DaRsin=zeros(Nk,Nl,Np,N);
  259. DbRcos=zeros(Nk,Nl,Np,N);
  260. DbRsin=zeros(Nk,Nl,Np,N);
  261. DcRcos=zeros(Nk,Nl,Np,N);
  262. DcRsin=zeros(Nk,Nl,Np,N);
  263. DdRcos=zeros(Nk,Nl,Np,N);
  264. DdRsin=zeros(Nk,Nl,Np,N);
  266. for ik=1:Nk
  267. kkk=k(ik);
  269. for il=1:Nl
  270. lll=l(il);
  271. for ip=1:Np
  272. pp=p(ip);
  273. idx=find(S_tr(:,1)==kkk & S_tr(:,2)==lll & S_tr(:,3)==pp);
  274. kk=kkk*2*pi/XRF;
  275. ll=2*pi*lll/XRF;
  277. xx=x_E(idx);
  278. yy=y_E(idx);
  280. Rcos(ik,il,ip,:)=pp*tensorprod( Lambda_tr(:,idx),cos( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ) , 2,1 );
  281. Rsin(ik,il,ip,:)=pp*tensorprod( Lambda_tr(:,idx),sin( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ) , 2,1);
  283. DaRcos(ik,il,ip,:)=-pp*tensorprod( Lambda_tr(:,idx), (kk*xx) .* sin( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  284. DaRsin(ik,il,ip,:)= pp*tensorprod( Lambda_tr(:,idx), (kk*xx) .* cos( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  286. DbRcos(ik,il,ip,:)=-pp*tensorprod( Lambda_tr(:,idx), (kk*yy) .* sin( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  287. DbRsin(ik,il,ip,:)= pp*tensorprod( Lambda_tr(:,idx), (kk*yy) .* cos( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  289. DcRcos(ik,il,ip,:)=-pp*tensorprod( Lambda_tr(:,idx), (ll*xx) .* sin( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  290. DcRsin(ik,il,ip,:)= pp*tensorprod( Lambda_tr(:,idx), (ll*xx) .* cos( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  292. DdRcos(ik,il,ip,:)=-pp*tensorprod( Lambda_tr(:,idx), (ll*yy) .* sin( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  293. DdRsin(ik,il,ip,:)= pp*tensorprod( Lambda_tr(:,idx), (ll*yy) .* cos( kk*(a*xx+b*yy)+ll*(c*xx+d*yy) ), 2,1);
  299. end
  300. end
  301. end
  304. end
  312. %%
  313. function [RFS,DRFS11,DRFS12,DRFS21,DRFS22]=RF_ESTSD(Lambda_tr,S_tr,x_E,y_E,RF_size,A)
  314. % This function computes the variables that are used to evaluate the objective function and its gradient
  315. [Rcos,Rsin,DaRcos,DaRsin,DbRcos,DbRsin,DcRcos,DcRsin,DdRcos,DdRsin]=reorganize(Lambda_tr,S_tr,x_E,y_E,A,RF_size);
  316. % [N,~]=size(Lambda_tr);
  317. Xmon=RF_size(1);
  318. Ymon=RF_size(2);
  319. k=unique(S_tr(:,1));
  320. l=unique(S_tr(:,2));
  321. p=unique(S_tr(:,3));
  324. Nk=length(k);
  325. Nl=length(l);
  326. % Np=length(p);
  330. casU=zeros(Nk,Nl,Xmon,Ymon);
  331. cas1U=zeros(Nk,Nl,Xmon,Ymon);
  333. Rcos1=squeeze(sum(Rcos,3));
  334. Rsin1=squeeze(sum(Rsin,3));
  336. DaRcos1=squeeze(sum(DaRcos,3));
  337. DaRsin1=squeeze(sum(DaRsin,3));
  339. DbRcos1=squeeze(sum(DbRcos,3));
  340. DbRsin1=squeeze(sum(DbRsin,3));
  343. DcRcos1=squeeze(sum(DcRcos,3));
  344. DcRsin1=squeeze(sum(DcRsin,3));
  346. DdRcos1=squeeze(sum(DdRcos,3));
  347. DdRsin1=squeeze(sum(DdRsin,3));
  349. for ik=1:Nk
  350. kk=k(ik);
  351. for il=1:Nl
  352. ll=l(il);
  353. U=KXLY(RF_size,kk,ll);
  354. casU(ik,il,:,:)=cas(U);
  355. cas1U(ik,il,:,:)=cas1(U);
  357. end
  359. end
  361. RFS=tensorprod(Rcos1,casU,[1,2],[1,2]) + tensorprod(Rsin1,cas1U,[1,2],[1,2]);
  362. DRFS11=tensorprod(DaRcos1,casU,[1,2],[1,2]) + tensorprod(DaRsin1,cas1U,[1,2],[1,2]);
  363. DRFS12=tensorprod(DbRcos1,casU,[1,2],[1,2]) + tensorprod(DbRsin1,cas1U,[1,2],[1,2]);
  364. DRFS21=tensorprod(DcRcos1,casU,[1,2],[1,2]) + tensorprod(DcRsin1,cas1U,[1,2],[1,2]);
  365. DRFS22=tensorprod(DdRcos1,casU,[1,2],[1,2]) + tensorprod(DdRsin1,cas1U,[1,2],[1,2]);
  370. end
  372. %%
  374. function RFS=RF_ESTS(Lambda_tr,S_tr,x_E,y_E,RF_size,A)
  376. [Rcos,Rsin,~,~,~,~,~,~,~,~]=reorganize(Lambda_tr,S_tr,x_E,y_E,A,RF_size);
  378. Xmon=RF_size(1);
  379. Ymon=RF_size(2);
  380. k=unique(S_tr(:,1));
  381. l=unique(S_tr(:,2));
  382. p=unique(S_tr(:,3));
  385. Nk=length(k);
  386. Nl=length(l);
  388. casU=zeros(Nk,Nl,Xmon,Ymon);
  389. cas1U=zeros(Nk,Nl,Xmon,Ymon);
  390. Rcos1=squeeze(sum(Rcos,3));
  391. Rsin1=squeeze(sum(Rsin,3));
  392. for ik=1:Nk
  393. kk=k(ik);
  394. for il=1:Nl
  395. ll=l(il);
  396. U=KXLY(RF_size,kk,ll);
  397. casU(ik,il,:,:)=cas(U);
  398. cas1U(ik,il,:,:)=cas1(U);
  400. end
  402. end
  406. RFS=tensorprod(Rcos1,casU,[1,2],[1,2]) + tensorprod(Rsin1,cas1U,[1,2],[1,2]);
  409. end
  413. %%
  416. function stop = outfun(x, optimValues, state)
  417. stop = false;
  418. A=[x(1), x(2) ; x(3) , x(4)];
  419. disp('A=')
  420. disp(A)
  421. drawnow
  422. end
  424. %%
  425. function [ELL,DELL]=EYE_CORRECT_COSTSD(X,x_E,y_E,S_tr,Lambda_tr,RF_size,mask)
  426. % This function returns
  427. % ELL = -E(A) where E(A) is our opjective function
  428. % DELL = the gradient
  430. A(1,1)=X(1);
  431. A(1,2)=X(2);
  432. A(2,1)=X(3);
  433. A(2,2)=X(4);
  435. [RFS,DRFS11,DRFS12,DRFS21,DRFS22]=RF_ESTSD(Lambda_tr,S_tr,x_E,y_E,RF_size,A);
  437. if(~isempty(mask))
  438. RFS=RFS.*mask;
  439. end
  440. DELL = -2 * [sum(RFS.*DRFS11,'all') , sum(RFS.*DRFS12,'all') , sum(RFS.*DRFS21,'all') , sum(RFS.*DRFS22,'all')]/(RF_size(1)*RF_size(2));
  441. ELL = -norm(RFS(:))^2 /(RF_size(1)*RF_size(2));
  443. end
  447. %%
  448. function y=cas1(x)
  450. y=sin(x)-cos(x);
  452. end
  454. %%
  456. function y=cas(x)
  458. y=sin(x)+cos(x);
  460. end
  462. %%
  463. function [mask,index,envelopes,Narea,N1]=getmasks(log,m,RF_size)
  464. % This function gets the data (log table), the threshold for masking 0<m<1
  465. % and RF size [length , height].
  467. % Returns:
  468. % masks,
  469. % index = cell indices sorted according to the naive RF sharpness,
  470. % envelopes = The Riesz Transforms
  471. % Narea = number of connected components after thresholding
  472. % N1 = Number of RFs with only one connected component
  473. rfs=RECEPTIVE_FIELDS(log,[],[],0,RF_size);
  475. [N,~,~]=size(rfs);
  476. mask=zeros(size(rfs));
  477. Cost=zeros(N,1);
  478. envelopes=zeros(size(rfs));
  479. Narea=zeros(N,1);
  480. for i=1:N
  481. E=rieszEnvelope(squeeze(rfs(i,:,:)));
  482. Eth=m*max(E,[],'all');
  483. % Eth=mean(E,'all')+1*std(E,[],'all');
  484. idx=E>Eth;
  487. CC = bwconncomp(idx);
  488. p = regionprops(CC,"Area");
  489. [maxArea,maxIdx] = max([p.Area]);
  490. [areas,IDX]=sort([p.Area],'descend');
  491. Narea(i)=CC.NumObjects;
  492. idx2= cc2bw(CC,ObjectsToKeep=maxIdx);
  493. idx2 = imdilate(idx2,strel("disk",8));
  494. ratio=maxArea/sum(areas);
  498. envelopes(i,:,:)=E;
  499. mask(i,:,:)=idx2;
  503. Cost(i)=ratio*sum(idx2.*squeeze(rfs(i,:,:).^2),'all');
  504. Cost(i)=Cost(i)/(100000^(Narea(i)-1));
  505. end
  508. ind=Narea==1;
  509. N1=sum(ind);
  511. [~,index]=sort(Cost,'descend');
  515. end
  519. %%
  523. function RFST=RECEPTIVE_FIELDS(DATA,select,A,flag,RF_size)
  525. % Inputs are:
  527. % DATA=log
  529. % "select" is an array indices of the neurons you want their receptive fields.
  530. % If select = [] all the neurons will be selected
  533. % Outputs are:
  534. % rfsb and rfsa
  535. % Receptive fields for neurons with and without corrections based on A.
  537. % flag=0 RFs are calculated for all time bins. falg=1 RFs are calculated for
  538. % the average 2:7 time bins.
  540. nstim = size(DATA,1);
  541. ncells = size(DATA.signal{1},2);
  542. if(isempty( select ))
  543. select=1:ncells;
  544. end
  546. if(isempty( A ))
  547. A=[0,0 ; 0,0];
  548. end
  550. x_E=DATA.xpos-mean(DATA.xpos,'omitnan');
  551. y_E=DATA.ypos-mean(DATA.ypos,'omitnan');
  553. idxnanx=isnan(x_E);
  554. idxnany=isnan(y_E);
  555. x_E(idxnanx)=0;
  556. y_E(idxnany)=0;
  558. S_tr=zeros(nstim,3);
  560. S_tr(:,1)=DATA.kx;
  561. S_tr(:,2)=DATA.ky;
  562. S_tr(:,3)=DATA.sign;
  564. Xmon=RF_size(1);
  565. Ymon=RF_size(2);
  567. LAMBDA=cat(3,DATA.signal{:});
  568. LAMBDA=LAMBDA(:,select,:);
  570. [T,N,~]=size(LAMBDA);
  573. if flag==1
  574. RFST=zeros(T,N,Xmon,Ymon);
  576. for t=1:T
  578. Lambda_tr=squeeze(LAMBDA(t,:,:));
  579. RFS=RF_ESTS(Lambda_tr,S_tr,x_E,y_E,RF_size,A);
  581. RFST(t,:,:,:)=RFS;
  583. end
  585. elseif(flag==0)
  586. Lambda_tr=squeeze(mean(LAMBDA(2:7,:,:),1));
  587. RFST=RF_ESTS(Lambda_tr,S_tr,x_E,y_E,RF_size,A);
  590. end
  592. end
  596. %%
  598. function E = rieszEnvelope(X)
  599. % RIESZENVELOPE Local envelope via the monogenic signal
  600. % E = rieszEnvelope(X) returns sqrt( X.^2 + R1.^2 + R2.^2 ),
  601. % where [R1,R2] are the Riesz transform of X.
  602. % compute Riesz components
  603. [M,N] = size(X);
  604. F = fft2(X);
  605. ux = ifftshift((-floor(N/2):ceil(N/2)-1)/N);
  606. uy = ifftshift((-floor(M/2):ceil(M/2)-1)/M).';
  607. [U,V] = meshgrid(ux,uy);
  608. K = sqrt(U.^2+V.^2)+eps;
  609. H1 = -1i*(U./K);
  610. H2 = -1i*(V./K);
  611. R1 = ifft2(F.*H1);
  612. R2 = ifft2(F.*H2);
  613. % envelope = magnitude of [X, R1, R2]
  614. E = sqrt( abs(X).^2 + abs(R1).^2 + abs(R2).^2 );
  615. end
  617. %%
  620. function [Rt,norms]=krnls(log)
  621. % This function gets the log data and returns the frequency domain RFs (fRFs) and their norms.
  623. % "norms" is a two dimensional array.
  624. % First index -> number of cells
  625. % Second index -> time
  628. % Rt are kernels for all conditions
  630. maxk = max(log.kx); % the maximum number of cycles per screen
  632. sz = size(log.signal{1}); % size of the signal [time bins x number of cells]
  634. Rt = zeros([2*maxk+1 2*maxk+1 sz]); % responses all
  636. for i = 1:size(log,1) % lets accumulate the resposnes and compute the mean at the end
  638. kx = log.kx(i)+maxk+1; % pick (kx,ky) for this trial
  639. ky = log.ky(i)+maxk+1; % shift the index so it starts at 1
  640. Rt(kx,ky,:,:) = squeeze(Rt(kx,ky,:,:)) + log.signal{i};
  642. end
  647. ncells = size(log.signal{1},2);
  649. % filter
  651. sigma = 3; % Little bit of smoothing the kernels
  652. norms=zeros(ncells,10);
  653. for n=1:ncells
  654. for i=1:10
  656. Rt(:,:,i,n) = imgaussfilt(squeeze(Rt(:,:,i,n)),sigma);
  657. krnl=squeeze(Rt(:,:,i,n));
  658. norms(n,i)=norm(krnl(:));
  662. end
  663. end
  666. end
  669. %% Extracting preferred frequency and orientation of the RFs
  672. function [poris,freqs,frfm]=freq_extract(Rt)
  673. % Tis function gets the fRFs as input and returns
  674. % poris = preferred orientations
  675. % freqs = preferred spatial frequencies
  676. % frfm = mean of the fRFs
  678. [~,~,N]=size(Rt);
  679. poris=zeros(N,1);
  680. freqs=zeros(N,1);
  681. frfm=zeros(N,1);
  684. t=1;
  685. for i=1:N
  687. krnl=squeeze(Rt(:,:,i));
  688. krnl=0.5*(krnl+flip(flip(krnl,1),2));
  689. [rs, cs]=find(krnl==max(krnl(:)));
  692. if(length(cs)>1)
  693. thetaa=(180/pi)*atan( (cs(2)-cs(1))/(rs(2)-rs(1)));
  694. nuu=0.5*sqrt( (cs(2)-cs(1))^2+(rs(2)-rs(1))^2);
  695. else
  696. thetaa=2*pi;
  697. nuu=0;
  698. end
  702. poris(i,t)=thetaa;
  703. freqs(i,t)=nuu;
  704. frfm(i,t)=mean(krnl(:));
  707. end
  711. end

eye_cal.m, under CC-BY-4.0 · at the source

Overview

Authors: S Amin Moosavi1, Elaine Tring1, Dario L Ringach2
ORCID iDs: S Amin Moosavi, Dario L Ringach
  1. Department of Neurobiology, David Geffen School of Medicine, University of California, Los Angeles, CA, USA
  2. Department of Neurobiology and Department of Psychology, David Geffen School of Medicine, University of California, Los Angeles, CA, USA
Institutions: University of California, Los Angeles (United States)
Journal: Journal of vision, volume 26, issue 8, article 9
Dates: received 26 September 2025; accepted 4 July 2026; published online 31 August 2026; in print August 2026
Type: Research article · Language: English
License: CC BY-NC-ND
Identifiers: DOI 10.1167/jov.26.8.9 · PMID 42671131 · PMCID PMC13533300 · OpenAlex W4414524836
Open access: gold, a free copy (OpenAlex)
Preprint: doi:10.1101/2025.09.24.678404
Status: code verified
Categories: behavior only (modality), mouse (organism), systems (subfield)
Methods: Single-unit activity, calcium imaging, Physiology & signal measures
Keywords: receptive fields, primary visual cortex, free viewing, subspace reverse correlation, simple cells
MeSH: Eye Movements*, Models, Neurological*, Retina*, Visual Cortex*, Visual Fields*, Animals, Mice, Photic Stimulation (* major topic)
Topic: Visual perception and processing mechanisms (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: NEI NIH HHS (R21 EY035064, R01 EY034488); NINDS NIH HHS (R01 NS116471, RF1 NS116471)
Citations: not cited yet (Europe PMC); 16 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

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

figshare 32257377

License: CC-BY-4.0
State: the link answers, verified on 26 September 2026
Evidence: files inventoried
Languages: MATLAB (1)
Size: 3 files, 1 script
Software Heritage: not checked
Found in: the acknowledgements
Holds: README
Not found: license file, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: Image Processing Toolbox (1 file), Optimization Toolbox (1 file)
Availability: 1 check, the latest on 26 September 2026: the link answers (HTTP 200)
  • 26 September 2026: the link answers (HTTP 200)
1 file
At the source: doi.org/10.6084/m9.figshare.32257377

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;
  • 1 script, 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 3, 28 September 2026

  • Authors: added S Amin Moosavi (0000-0002-8862-789X); Dario L Ringach (0000-0001-6439-334X); removed S Amin Moosavi; Dario L Ringach

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 3 authors, 5 keywords, 8 MeSH terms, 2 funders, 12 references.

Cite

This paper

Moosavi, S. A., Tring, E., & Ringach, D. L. (2026). Subspace reverse-correlation estimation of receptive fields during free viewing. Journal of vision, 26(8), 9. https://doi.org/10.1167/jov.26.8.9

BibTeX

@article{moosavi2026subspace,
author = {Moosavi, S Amin and Tring, Elaine and Ringach, Dario L},
title = {{Subspace reverse-correlation estimation of receptive fields during free viewing}},
journal = {Journal of vision},
year = {2026},
month = aug,
volume = {26},
number = {8},
pages = {9},
publisher = {Association for Research in Vision and Ophthalmology},
issn = {1534-7362},
doi = {10.1167/jov.26.8.9},
url = {https://doi.org/10.1167/jov.26.8.9},
pmid = {42671131},
pmcid = {PMC13533300}
}

RIS

TY - JOUR
AU - Moosavi, S Amin
AU - Tring, Elaine
AU - Ringach, Dario L
TI - Subspace reverse-correlation estimation of receptive fields during free viewing
T2 - Journal of vision
J2 - J Vis
PY - 2026
DA - 2026/08/01
VL - 26
IS - 8
SP - 9
SN - 1534-7362
PB - Association for Research in Vision and Ophthalmology
DO - 10.1167/jov.26.8.9
UR - https://doi.org/10.1167/jov.26.8.9
LA - en
ER -

CSL-JSON

{
"id": "10.1167/jov.26.8.9",
"type": "article-journal",
"title": "Subspace reverse-correlation estimation of receptive fields during free viewing",
"container-title": "Journal of vision",
"author": [
{
"family": "Moosavi",
"given": "S Amin"
},
{
"family": "Tring",
"given": "Elaine"
},
{
"family": "Ringach",
"given": "Dario L"
}
],
"container-title-short": "J Vis",
"volume": "26",
"issue": "8",
"page": "9",
"DOI": "10.1167/jov.26.8.9",
"PMID": "42671131",
"PMCID": "PMC13533300",
"ISSN": "1534-7362",
"publisher": "Association for Research in Vision and Ophthalmology",
"URL": "https://doi.org/10.1167/jov.26.8.9",
"language": "en",
"issued": {
"date-parts": [
[
2026,
8,
1
]
]
}
}

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:10.1126/sciadv.adz9632 [code]
Population coding under the scale invariance of high-dimensional noise.
Journal: Science advances
In common: systems, mouse, author S Amin Moosavi
[2] doi:10.1016/j.crmeth.2026.101421 [code]
EthoPy provides an accessible platform for reproducible behavioral neuroscience.
Journal: Cell reports methods
In common: behavior only, mouse, 3 references
[3] doi:10.1038/s41467-026-75347-4 [code]
Sleep reveals dynamics integrating and segregating movement and stimulus representations in V1.
Journal: Nature communications
In common: Optimization Toolbox, Image Processing Toolbox, systems, mouse, 1 reference
[4] doi:10.1016/j.celrep.2026.117646 [code]
Medial entorhinal-hippocampal desynchronization parallels the emergence of memory impairment in a mouse model of Alzheimer's disease pathology.
Journal: Cell reports
In common: Optimization Toolbox, Image Processing Toolbox, systems, mouse, 1 reference
[5] doi:10.1016/j.neuron.2026.03.034 [code]
Dentate gyrus interneurons modulate winner-take-all network dynamics in freely behaving mice.
Journal: Neuron
In common: Optimization Toolbox, Image Processing Toolbox, systems, mouse, 1 reference
[6] doi:10.1523/jneurosci.2001-25.2026 [code]
Dynamics of Dentate Gyrus Place Cells and Dentate Spikes during Spatial and Nonspatial Changes in Environments.
Journal: The Journal of neuroscience : the official journal of the Society for Neuroscience
In common: Optimization Toolbox, Image Processing Toolbox, systems, 1 reference
[7] doi:10.1038/s41467-026-76581-6 [code]
Thalamocortical bursts encode reward contingencies and drive associative learning.
Journal: Nature communications
In common: Image Processing Toolbox, systems, mouse, 2 references
[8] doi:10.1523/eneuro.0417-25.2026 [code]
Learning and Motivation State Fluctuations from Motoric and Neurophysiologic Metrics during a Somatosensory Task in Mice.
Journal: eNeuro
In common: Optimization Toolbox, Image Processing Toolbox, behavior only, mouse
[9] doi:10.1038/s41593-026-02357-2 [code]
Experience reorganizes content-specific memory traces in macaques.
Journal: Nature neuroscience
In common: Optimization Toolbox, Image Processing Toolbox, 1 reference
[10] doi:10.1038/s41593-026-02350-9 [code]
Probing inter-areal computations with a two-photon holographic mesoscope.
Journal: Nature neuroscience
In common: Optimization Toolbox, Image Processing Toolbox, systems, mouse

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.

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: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

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