Error: ()-indexing must appear last in an index expression.

Hello,

I am currently stuck on a homework problem and could use some help.

I am supposed to write a function that takes as input 2 variables and outputs parameter estimates for the slope and bias of a cumulative Gaussian psychometric curve to fit to the data.

So far I have:

function [bias, slope] = psychometricFit(x, y)
p = normcdf(x, y, 1)
Gaussian = @(p,x) (0.5*erfc(x-p(1)/(p(2)(sqrt(2)))))
coefEst = nlinfit(x,y,Gaussian,[1 1])
end

I keep getting an error that says:

Error: File: psychometricFit.m Line: 3 Column: 37
()-indexing must appear last in an index expression.

I'm not sure how to fix this or "index" it properly.

Thank you

[Stephen23]

As well as the parentheses issue explained by Guillaume below, there is another problem: the third line defines p as an input to an anonymous function, and so the p on the second line

 p = normcdf(x, y, 1)

is never used.

Answer 1

[Guillaume]

In your anonymous function what is

p(2)(sqrt(2))

meant to do? It's clearly not valid matlab. Perhaps you are missing an operator.

[Taylor C]

I was trying to follow along with a code that we were shown in class which was y = mx + b

modelFun = @(p,x)*x+p(2)

[Taylor C]

The cumulative Gaussain function is

f(x) = 1/2*erf((x-mu)/(sigma(sqrt(2)))

my teacher gave us a hint that mu is the parameter for the bias and sigma is the parameter for the slope

[Guillaume]

modelFun = @(p,x)*something

is not valid matlab either.

In

f(x) = 1/2*erf((x-mu)/(sigma(sqrt(2)))

sigma(sqrt(2)) is not even mathematically valid notation (unless sigma is a function, which it clearly isn't if it's the value of the standard deviation). The actual definition of the CDF of the normal distribution is:

f(x) = 1/2*erf((x-mu)/(sigma*sqrt(2))

As hinted in my answer, you are missing an operator.

See also, the comment by Stephen about p.