How does mscohere calculate output frequency cord when no input frequency cord is given??

Hi,

I am trying to work with mscohere for the first time and I don't understand how does it calculates the output frequqncy when no input frequency cord is given. Here is a simple example through which I am trying to understand this.

%% Creat a Step Fuction
t=1:0.01:3;
h=ones(1,201); 
h(1:101)=0;
figure; plot(t,h,'.-k','linewidth',1.5,'markersize',12);

For simplicity I calculate correlation of the function with itself.

%% Calculate Coherence function
freq=1./t;
fs=1/0.01;
[Cxy, freqs] = mscohere(h,h, [], [],freq, fs);
figure;
plot(freqs, Cxy,'.-r', 'linewidth',2, 'markersize',18);

then I get same frequency cord as input freqency, which makes sens that it is calculating correlation at each point of the given frequency. But if I use the following:

[Cxy, freqs] = mscohere(h,h, [], [], fs);

I get 51 values in output frequency and values are completely different then previos case. I do not understand, how it calculates when no input frequency is given. I think it is related to default value of the hamming window but I don't understand it clearly. How does it choose those particular frequencies?

Could someone please help me to understand this?