Visual perceptual learning (VPL) refers to a long-term enhancement of visual task performance as a result of
visual experience. The global aim of the current proposal is to investigate VPL of natural scenes/stimuli (NS).
VPL is an important tool that can be used to better understand visual plasticity and to improve and/or restore
damaged, declining or degraded vision. In almost all studies of VPL, artificial and unnatural stimuli (US)
Gabor patches VPL of US has been found to be restricted by at least following two
, have been used for training.
factors. First, task-relevant VPL (R-VPL) and task-irrelevant VPL (I-VPL) of US tend to be specific for the trained
are regarded as serious limitations of VPL in clinical applications that require generalized effects on
suprathreshold NS in everyday life.
location. Second, if US is suprathreshold, I-VPL of the US does not occur. These restrictions on VPL
However, here we raise the possibility that these restrictions are not
necessarily applied to VPL of NS. We plan to test this possibility by systematically comparing the degree of
restrictions on VPL of NS with that on VPL of US. If smaller amounts of restrictions on VPL of NS indeed occur,
we will further test whether the same results occur with each of the following 4 different characteristics of NS: (1)
vertical or horizonal orientation dominance, (2) specific spatial frequency distributions (1/f, spatial correlation),
(3) Portilla & Simoncelli statistics that include correlations of different orientation and SF channels as well as
these channels themselves from original NS images and (4) categorized items or categorical organization.
Specific Aim 1 will test how different characteristics of NS reduce restrictions on I-VPL.
A viewer in a natural environment is exposed to a multitude of suprathreshold task-irrelevant features. Therefore,
it is important to examine whether and how VPL of NS occurs. We will test the following hypotheses for VPL of
discrimination and detection tasks. Hypothesis (H)1-1: I-VPL of orientation for suprathreshold NS occurs,
whereas that for suprathreshold US does not. H1-2: I-VPL of orientation for suprathreshold US occurs if the US
have each of the above-mentioned 4 characteristics of NS. H1-3: I-VPL of orientation for NS transfers more to
untrained orientations and/or locations than I-VPL of orientation for US. H1-4: The transfer of I-VPL of orientation
for US occurs if the US have each of the above-mentioned 4 NS characteristics.
Specific Aim 2 will test how different characteristics of NS reduce restrictions on R-VPL.
R-VPL of NS is crucial because a viewer conducts a goal-oriented task in everyday life and needs to improve
task performance regarding NS. We will examine whether R-VPL of NS is more generalized to untrained
features (orientations), locations and tasks than VPL of US by testing the following hypotheses. H2-1: R-VPL
of orientation for NS transfers to untrained orientations, locations and/or tasks. H2-2: The transfer of R-VPL of
orientation for US occurs if the US have each of the above-mentioned 4 characteristics of NS.