Thought Experiments (TEs) are reasoning processes that are based on 'result
s' of an experiment carried out in thought. What is the validity of an expe
riment - that has not been actually executed - for knowledge about the phys
ical world! What are the features that make it distinctive and how do we in
tegrate it into learning environments to support such thought processes! Th
is study suggests that a thought experiment draws on three epistemological
resources: conceptual-logical inferences, visual imagery and bodily-motor e
xperience. We start by stating how students' TEs are related to recent rese
arch on learning science and then proceed to describe the nature of TEs. Th
e central part of the paper deals with cognitive theories and empirical exa
mples of visual imagery and bodily imagery. It also deals with how these en
able implicit knowledge about the world to be retrieved. Students may have,
but are not aware of, such knowledge for it is hidden when learning is onl
y based on format representations. We show that imagination is structured,
goal-oriented, based on prior experiential imagery and internally coherent.
Students can, for example, mentally rotate objects at constant velocity. S
tudents can 'zoom in and out' to inspect imaginary situations, transfer obj
ects, predict paths of imaginary moving objects and imagine the impact of f
orces on mechanical systems. We show thar the TEs are powerful because of t
hese capabilities. We further claim that these are not exploited by school
learning environments and offer a first step towards understanding imagery
in science learning.