Extrapolating values from a glm fit

Dear R-help,

I have fitted a glm logistic function to dichotomous forced choices
responses varying according to time interval between two stimulus. x  
values
are time separation in miliseconds, and the y values are proportion
responses for one of the stimulus. Now I am trying to extrapolate x  
values
for the y value (proportion) at .25, .5, and .75. I have tried several
predict parameters, and they don't appear to be working. Is this  
correct use
and understanding of the predict function? It would be nice to know  
the
parameters for the glm best fit, but all I really need are the  
extrapolated
x values for those proportions. Thank you for your help. Here is the  
code:

x <-
c(-283.9, -267.2, -250.5, -233.8, -217.1, -200.4, -183.7, -167,
-150.3, -133.6, -116.9, -100.2, -83.5, -66.8, -50.1, -33.4, -16.7,
16.7, 33.4, 50.1, 66.8, 83.5, 100.2, 116.9, 133.6, 150.3, 167,
183.7, 200.4, 217.1, 233.8, 250.5, 267.2, 283.9)

y <-
c(0, 0.333333333333333, 0, 0, 0, 0, 0, 0, 0, 0.333333333333333,
0, 0.133333333333333, 0.238095238095238, 0.527777777777778,
0.566666666666667,
0.845238095238095, 0.55, 1, 0.888888888888889, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 0.5)

weight <-
c(1, 3, 2, 5, 4, 4, 3, 5, 5, 4, 5, 11, 22, 11, 15, 16, 11, 7,
14, 10, 16, 19, 11, 5, 4, 5, 6, 9, 4, 2, 5, 5, 2, 2)

mylogit <- glm(y~x,weights=weight, family = binomial)

# now I try plotting the predicted value, and it looks like a good  
fit,
hopefully I can access what the glm is doing

ypred <- predict(mylogit,newdata=as.data.frame(x),type="response")
plot(x, ypred,type="l")
points(x,y)

# so I try to predict the x value when y (proportion) is at .5, but
something is wrong..

predict(mylogit,newdata=as.data.frame(0.5))

	[[alternative HTML version deleted]]