Error in model function using fitnlm

Hello everyone!

I'm having some trouble with a code I'm working on. I need to make non linear regression of some datas I have stored using different model functions but I can't make it work.

The central part of the code is

load('Dati_3-08.dat'); 

f1=@(x,a) a(1)+a(2).*(x-1.1);
f2=@(x,a) a(1)+a(2).*(x-1.1)+a(3).*(x-1.1).^2;
f3=@(x,a) a(1)+a(2).*(x-1.1)+a(3).*(x-1.1).^2+a(4).*(x-1.1).^3;
f4=@(x,a) a(1).*exp(a(2).*(x-1.1).^2);

a1=randn(2,1);
a2=randn(3,1);
a3=randn(4,1);
a4=randn(2,1);

r1=fitnlm(Dati_3_08(:,1),Dati_3_08(:,2),f1,a1);
r2=fitnlm(Dati_3_08(:,1),Dati_3_08(:,2),f2,a2);
r3=fitnlm(Dati_3_08(:,1),Dati_3_08(:,2),f3,a3);
r4=fitnlm(Dati_3_08(:,1),Dati_3_08(:,2),f4,a4);

where Dati_3_08 is a 18x2 matrix.

The error I keep getting is: "MODELFUN must be a function that returns a vector of fitted values the same size as Y (18-by-1). The model function you provided returned a result that was 2-by-1."

I had the doubt the error was due to f1 being linear, but the same thing happen with f2 and the others model functions. What am I doing wrong?

Also, with f1 being linear should I change fitnlm (since it's for non linear regression) in a linear regression function or should it still work?