Cognitive and Neuroscientific Models of Memory

Evolutionary Bias and Survival Stimuli

Evolutionary Benefit: Stimuli that capture attention (e.g., snakes and spiders) do so because it is evolutionarily beneficial for survival. This is viewed as a survival mechanism where specific stimuli facilitate attention automatically.
Limitations of Current Research: A primary limitation in studying this is the limited range of stimuli. While researchers often use snakes and spiders as examples of evolutionary threats, these represent a relatively small portion of the total range of evolutionary threats that could have impacted human survival.

Neural Circuitry of Emotional Memories

Key Brain Regions: The amygdala and the hippocampus are the primary regions that interact to support the decoding of emotional memories.
Predicting Recall Performance:
- Activity in the amygdala complex is a significant predictor of the recall of emotional details (e.g., from emotional films).
- Specifically, the Glucose Metabolic Rate (GMR) in the amygdala is used as a metric.
- Positive Correlation: A higher GMR in the amygdala predicts a higher level of recall for emotional details.

Bridging Cognitive and Neuroscientific Models

Historical Context: Memory was studied by cognitive psychologists and philosophers of mind (e.g., the short-term memory model with the phonological loop or the declarative vs. non-declarative distinction) long before neuroscience became a distinct field. Even behaviorism relied on memory ideas, though they avoided the term due to their rejection of internal processes.
Modern Approach: This course aims to link classic cognitive psychology effects (behavioral observations) with modern neuroscience (neural mechanisms) to understand why these behaviors occur without focusing solely on isolated cell biology.

The Spacing Effect

Definition: Memory and recall are consistently better when learning sessions are repeated and spaced out over time (spaced learning) compared to when they are grouped into a single, continuous block (massed learning).
Discovery: This effect was identified over $100$ years ago by Hermann Ebbinghaus in Germany.
Magnitude vs. Spacing: The benefit is not about the total amount of time spent; it is about the act of spacing. For example, three $1$ -hour study blocks will produce better recall than one single $3$ -hour block.
Breadth of Application: The spacing effect applies to all forms of learning, including classical conditioning, operant conditioning, and any stimulus-response exposure. Spacing leads to faster conditioning and more robust learning.

Applied Spaced Repetition: The Leitner System

The Leitner System: A practical application of spaced repetition using flashcards.
Procedure:
- Flashcards are sorted into five boxes (or piles) based on perceived difficulty.
- Promotion: If a card is answered correctly, it moves to the next box to the right (e.g., from Box $1$ to Box $2$ ).
- Demotion: If a card is answered incorrectly, it is moved all the way back to Box 1, regardless of where it was.
Review Frequency: Box $1$ is reviewed every day, Box $2$ every two days, Box $3$ twice a week, and so on. This system ensures the hardest material is seen most frequently while easy material is spaced further apart.

Neural Mechanisms: The Molecular Refractory Period Hypothesis

Core Concept: Learning is underpinned by Long-Term Potentiation (LTP), the strengthening of synaptic connections. The Molecular Refractory Period Hypothesis suggests that LTP has a recovery period where it is weaker immediately after it has occurred.
Why Spacing Works: Rapid, repeated stimulation (massed training) cannot trigger multiple rounds of LTP effectively because the system hasn't recovered. Spacing allows the neural mechanism to "reset," allowing for a second round of potentiation and a stronger overall memory trace.
Two Sub-Hypotheses for LTP Changes:
- Volume Explanation: LTP causes postsynaptic dendritic spines to increase in volume. This requires time and resources. In massed training, they grow once and cannot grow further immediately. Spaced training allow for sequential growth phases.
- Recruitment Explanation: Only some spines grow during the first exposure, while others are primed. Primed spines do not change immediately. In massed training, the second exposure occurs while they are still just primed. In spaced training, enough time passes for the primed spines to be ready to undergo LTP and increase in volume upon the second exposure.

Levels of Processing (LOP) Model

Authors: Craig and Lockhart ( $1972$ ).
Theory: The depth of cognitive processing at the time of encoding determines the success of retrieval.
Three Identified Typical Levels:
- Structural (Orthographic): Focusing on the visual or spelling qualities of a word (e.g., "Does the word start with a vowel?"). This is the shallowest level and leads to the poorest recall.
- Phonemic (Auditory): Focusing on the sound of the word (e.g., "Does the word rhyme with towel?"). This is a mid-level of processing.
- Semantic: Focusing on the meaning and categorization of the word (e.g., "Is this a type of vehicle?"). This is the deepest level and leads to the best recall.
Neuroscientific Finding: Processing differences are qualitative, not quantitative. Shallow and deep processing do not just use "more" or "less" of the same brain region; they rely on entirely different brain regions. Shallow processing often activates posterior regions (parietal/occipital), while deep processing activates frontal regions.

Transfer Appropriate Processing (TAP)

Core Argument: Recall performance is determined by the match between the encoding process and the recall process, rather than the depth of encoding alone.
Evidence: Semantic encoding is only "best" because most recall tasks are semantic. If a recall task required structural information (e.g., identifying vowels), structural encoding would actually outperform semantic encoding.
Mechanism: Encoding with one specific brain region makes it easier to recall if the same brain region is activated for the retrieval task. A mismatch in regions leads to poor performance.

The Serial Position Effect

Observation: The order of items in a list influences how well they are remembered, typically forming a U-shaped curve.
Primacy Effect: Better recall for items at the start of a list. These items are thought to be encoded more effectively due to novelty or increased attention.
Recency Effect: Better recall for items at the end of a list. These are thought to be still held in short-term or working memory.
Interference: If a distractor or delay is introduced before recall, the recency effect is abolished, but the primacy effect remains strong.

Models of Serial Position

Single Store Model: Suggests one memory system explains the effect. Primacy happens because early items are more "discriminable" or receive more attention; recency happens because items are "temporally closer" to the recall event.
Dual Store Model: Suggests two distinct memory systems. Early items are encoded into a long-term/longer-duration store. Late items sit at the forefront of the working memory (e.g., the phonological loop). Middle items fail to be encoded in either effectively.
Neuroscience and Lesion Studies:
- Human fMRI: Early items (primacy) show significantly greater medial temporal lobe (MTL) and hippocampal activation. Late items do not show this hippocampal activation.
- Rat Studies: Rats trained on mazes show a serial position effect. A bilateral hippocampal lesion in rats eliminates the primacy effect but leaves the recency effect intact, suggesting the hippocampus is specifically required for the primacy component.

The DRM (Deese-Roeder-McDermott) Paradigm

Effect: This paradigm is designed to induce false memories. Participants are presented with a list of words (e.g., nurse, hospital, medicine, dentist) that are all semantically related to a lure word or critical word (e.g., doctor) that is not actually present.
Result: Participants frequently "recall" the lure word with high levels of confidence.
Theories of DRM:
- Activation Monitoring Theory: Each presented word "pings" its semantic associations. Because the lure word is at the center of all these associations, it receives repeated activation during encoding and is retrieved as part of the list.
- Fuzzy Trace Theory: Proposes two types of memory traces: Gist (the general meaning) and Verbatim (exact episodic details). False memories occur because the "gist" of the list includes the lure word.
Semantic vs. Episodic Conflict: Higher semantic memory ability predicts a higher likelihood of producing the false lure word. Higher episodic memory ability (verbatim memory) predicts a lower likelihood, as it helps the brain recognize that the specific word did not appear on the slide.

Questions & Discussion

Question: What is LTP?
Response: Long Term Potentiation. It refers to the alteration of the strength of the synaptic connection between a pre-synaptic neuron and a post-synaptic neuron. It is the physical change that underpins learning and memory.
Question: Can you repeat what the quantitative vs. qualitative difference for Levels of Processing was?
Response: Quantitative would be if one region activated less for shallow and more for deep. Qualitative means they rely on fully different, distinct brain regions for encoding and recall.
Question: What was the study in Psychology 2050 (Learning and Cognition) regarding this?
Response: Students noted that while the Serial Position Effect is taught, the Single vs. Dual Store models may not be covered in the same way in that course.
Question: What is the worksheet hint?
Response: The spacing effect is on the worksheet. Specifically, the relationship between spaced learning being better than massed learning and how this is explained by the molecular refractory period of LTP.
Question: Can you summarize the spacing effect for the class?
Response: Spaced learning is better than massed learning. The popular theory is the Molecular Refractory Period, which says the brain/LTP needs time to recover to be effective again.