Outline and evaluate the contemporary study of Steyvers and Hemmer (2012).
The aim of Steyvers and Hemmer’s study was to investigate the interaction between episodic and
prior knowledge in naturalistic environments. They argued that previous studies for reconstructive
memory had poor ecological validity, since they were all conducted in laboratory environments.
They suggested that these studies deliberately induced errors in recall, thus creating a ‘false mem-
ory’. In summary, Steyvers and Hemmer wanted to see how prior knowledge (semantic memory)
was used to reconstruct memory for photographs of normal everyday settings (episodic recall), such
as hotel, kitchen, and office.
The procedure of this experiment consisted of two sections - the Expectation Test and the Percep-
tion Test. The main reason for the Expectation test was to measure the frequency of all the objects
the participants would name. During this test, twenty-two random participants were chosen from the
University of California, and were told to recall the objects they would expect to find (thus the ‘ex-
pectation’ test) in five naturalistic scenes (office, kitchen, hotel, urban setting, and dining room).
They were told to enter their responses on a computer for at least one minute per scene. For the Per-
ception Test, a separate group of twenty-five participants were shown twenty-five images of the
five scenes, and were asked to recall all the objects they could see. The importance of this initial
testing was to ensure that objects were not overlooked just because they were not perceptible in the
image (aka. the objects recalled were from semantic memory - they were deliberately taken out of
the picture). The results of this initial testing suggested that people had goof prior knowledge (se-
mantic memory) of each scene, increasing the ecological validity of their experiment through the
use of naturalistic environments.
After the procedure, Steyvers and Hemmer continued to further increase the reliability of their study
through the experimental memory condition. Using the same experimental pool (or University of
California students) 49 participants who had not taken part in any of the two initial experiments
were randomly selected. In this experiment, ten of the stimulus images from the Perception Test
were used (two of each scene which resulted to the most objects named), and from the two sets, one
from each scene, were formed. Creating two groups from the 49 participants, one group were shown
images for 2 seconds, whilst the other, 10 seconds. There were four possible trial time orderings,
and the participants were randomly allocated to one of the time orderings.
The results from the experiment showed that incorrect recall of highly probable objects was 9% and
incorrect recall of low probability objects was 18%. These results ran in contrast to previous re-
search, which suggested that high error rates would happen for high probability objects. However,
the low error rate for high probability objects was not surprising as they were likely to be seen in
unmanipulated naturalistic scenes.
What is good about their study was that they were able to show a high ecological validity for their
results in comparison to the lab results from previous studies of reconstructive memory. However,
although they were able to increase the ecological validity of their study, they still conducted their
experiment in a lab, unable to expose their participants to real environments. Nevertheless, they
were able to control many potential factors to their experiment. For example they controlled the
number of images the participants were shown, preventing interference from a perviously viewed
scene.
The aim of Steyvers and Hemmer’s study was to investigate the interaction between episodic and
prior knowledge in naturalistic environments. They argued that previous studies for reconstructive
memory had poor ecological validity, since they were all conducted in laboratory environments.
They suggested that these studies deliberately induced errors in recall, thus creating a ‘false mem-
ory’. In summary, Steyvers and Hemmer wanted to see how prior knowledge (semantic memory)
was used to reconstruct memory for photographs of normal everyday settings (episodic recall), such
as hotel, kitchen, and office.
The procedure of this experiment consisted of two sections - the Expectation Test and the Percep-
tion Test. The main reason for the Expectation test was to measure the frequency of all the objects
the participants would name. During this test, twenty-two random participants were chosen from the
University of California, and were told to recall the objects they would expect to find (thus the ‘ex-
pectation’ test) in five naturalistic scenes (office, kitchen, hotel, urban setting, and dining room).
They were told to enter their responses on a computer for at least one minute per scene. For the Per-
ception Test, a separate group of twenty-five participants were shown twenty-five images of the
five scenes, and were asked to recall all the objects they could see. The importance of this initial
testing was to ensure that objects were not overlooked just because they were not perceptible in the
image (aka. the objects recalled were from semantic memory - they were deliberately taken out of
the picture). The results of this initial testing suggested that people had goof prior knowledge (se-
mantic memory) of each scene, increasing the ecological validity of their experiment through the
use of naturalistic environments.
After the procedure, Steyvers and Hemmer continued to further increase the reliability of their study
through the experimental memory condition. Using the same experimental pool (or University of
California students) 49 participants who had not taken part in any of the two initial experiments
were randomly selected. In this experiment, ten of the stimulus images from the Perception Test
were used (two of each scene which resulted to the most objects named), and from the two sets, one
from each scene, were formed. Creating two groups from the 49 participants, one group were shown
images for 2 seconds, whilst the other, 10 seconds. There were four possible trial time orderings,
and the participants were randomly allocated to one of the time orderings.
The results from the experiment showed that incorrect recall of highly probable objects was 9% and
incorrect recall of low probability objects was 18%. These results ran in contrast to previous re-
search, which suggested that high error rates would happen for high probability objects. However,
the low error rate for high probability objects was not surprising as they were likely to be seen in
unmanipulated naturalistic scenes.
What is good about their study was that they were able to show a high ecological validity for their
results in comparison to the lab results from previous studies of reconstructive memory. However,
although they were able to increase the ecological validity of their study, they still conducted their
experiment in a lab, unable to expose their participants to real environments. Nevertheless, they
were able to control many potential factors to their experiment. For example they controlled the
number of images the participants were shown, preventing interference from a perviously viewed
scene.
