The thesis of relation between observation and theory is one of the basic important issues in philosophy of science and scientific epistemology. However

the mechanistic processes of theory- ladenness of observation have rarely been discussed. Current research in cognitive science on thought processes provides powerful analytical tools and empirical support for this problem. In the light of the perception-based knowledge representation of Barsolou

this paper attempts to give a representation-based explanation for theory- laden mechanism in virtue of constraints on production and explanation of representation. If representational mechanism in observation is not an absolutely subjective but a constraint construction

it will give the reliability of observation a rational support.