How can we possibly explain
the inaccuracy of observers’
reports about distance if, as I
have just argued, the results
for size are based on the
assumption that the distance
change is perceived
appropriately? The answer, I
believe, is similar to the one
advanced in Chapter 2 to
explain a paradox about the 
moon illusion: Observers 
perceive the horizon moon to 
be larger because they perceive 
it to be farther away, yet 
precisely because they do 
perceive it to be larger, they 
conclude that it must appear 
closer, since they have learned 
that size varies with distance. 
In the convergence 
experiment, observers are also 
faced with two conflicting 
sources of information to 
which they must respond, and 
they also make a deduction 
about distance based not on a 
genuine perceptual cue but on 
apparent size. With an increase 
in convergence, the figure 
appears to shrink in size; 
observers are thus inclined to 
think, in response to the 
experimenter's question about 
distance change, "It is getting 
smaller, so it must be moving 
farther away." But if they use 
convergence as a direct cue to 
distance, they are inclined to 
give the opposite response. 
Information is available that 
the convergence of the eyes is 
increasing. Therefore, the 
perceptual system can infer 
that the eyes must be focusing 
on something that is coming 
closer. Observers thus face a 
conflict when it comes to 
responding about distance, but 
not when it comes to 
responding about size. That is 
why reports about size can 
sometimes be a more accurate, 
less-contaminated measure of 
the effectiveness of distance 
cues than are distance reports 
themselves. It is also why, in 
the convergence experiment, 
reports about size are more 
consistent than reports about 
distance.