Psychology Long-Term Memory Studies

Study Notes

An approach to study LTM compares neural activation during the presentation of novel stimuli with the same during familiar stimuli.
Stimulus novelty likely results in more robust encoding mechanisms being automatically recruited.
The research focuses on encoding, not recollection.

Researchers used a paradigm of presenting novel and familiar stimuli to study encoding in LTM.
They were interested in whether encoding processes differed between novel and familiar stimuli.
The experiment used a paradigm to test encoding process, but the researchers did not specifically probe recollection.

PDP models are built to test hypotheses and simulate human behavior during a particular task.
PDP models can help answer the question, "Could a system based on interconnected neurons accomplish object recognition?"
It tests whether PDP systems are able to accomplish certain tasks, and if so, whether this implies that the human brain operates similarly.

A PDP model was created to simulate human facial recognition judgments.
The model was tested to see if it could match human behavior.
This model demonstrates that even if a system can perform a task, it doesn't necessarily explain how humans do that task.

PDP models can be used to test more specific hypotheses, such as the effects of simulated neurological damage on the system's output.
PDP models can simulate neurological damage (e.g., stroke) to understand how it affects performance.
Simulated neurological damage in a PDP model can produce errors that mimic those exhibited by patients with semantic dementia.

PDP models show graceful degradation, meaning selective damage to certain units doesn't necessarily result in a complete system breakdown.
This is similar to how neurological damage in humans can have varying effects depending on the damage's extent.
This contrasts with other kinds of systems where damage can result in complete failure (for example, damage to an AC adaptor).

The model's graceful degradation contrasts with the hypothesis that knowledge is stored in specific neurons or groups of neurons, which would lead to more selective losses.
The "selective-loss" hypothesis implies specific damage would lead to specific knowledge loss, but neurological damage often results in more generalized effects.
However, some cases suggest specific behavioral consequences from neurological damage, such as visual agnosia that only affects musical instruments.

PDP models challenge the idea that knowledge is stored in an "amodal" form, independent of sensory modalities.
Instead, they suggest that meaning emerges from the interaction of all parts of the system, not specific locations.
This is related to the Gestalt principle that "The whole is more than the sum of its parts."

Early work on long-term memory using imaging highlighted challenges in design and analysis.
One challenge was using block designs, which were typical at the time, in the experiment.

The study by Petersen et al. (1988) compared activation during word blocks with fixation blocks using subtractive logic.
The study used a block design comparing activation during word blocks with fixation blocks.
There are methodological issues with this study, such as unclear participant activity during fixation blocks.