Models of Memory

The multi-store memory model

• Memory is a cognitive process used to encode, store and retrieve information.

• The multi-store memory model was proposed by Atkinson and Shiffrin in 1968. In this model, human memory is said to consist of three separate components:

  • sensory memory

  • short-term memory store

  • long-term memory store.

• Each of these components is characterized by a specific duration (for how long the store is able to hold information) and capacity (how many units of information it can hold).

• In order for information to move to the next memory store, certain conditions have to be met.

  Sensory Memory Store

• The sensory memory store does not process information.

• Its function is to detect information and hold it until it is either transferred further into the short-term memory store or lost.

• The capacity of sensory memory is only limited by our perception, for example, iconic memory can keep everything that enters our visual field and echoic memory can hold everything that we acoustically perceive at any moment.

• However, the duration of sensory memory is short. Traces in iconic memory decay after 1 second of inattention, while traces in echoic memory can decay after 2–5 seconds.

• The condition that has to be met for information to transfer from sensory memory to short-term memory is attention. If a unit of information is attended to, it does not decay but moves to the next memory store. Conversely, if it does not catch our attention, it fades away.

Short-term memory Store

• As information enters the short-term memory (STM) store, it can undergo some primitive transformations.

• The capacity of short-term memory is 7±2 chunks of information.

• The duration of short-term memory is no longer than 30 seconds. If the information is left unattended, the trace fades away in this period of time.

• The condition for increasing the duration of short-term memory and transferring information into the third store (long-term memory) is rehearsal. If we rehearse information, it stays in the STM longer, and eventually, the trace gets consolidated and the information enters the long-term memory store.

Long-term memory store

• Long-term memory (LTM) is described as a place for storing large amounts of information for indefinite periods of time.

• The current estimate of the capacity of LTM is that it is potentially virtually unlimited.

• Although the capacity of LTM is potentially unlimited, not all information that is stored in LTM is easily retrievable.

• It is not storing but retrieving information from memory that may be problematic.

• Similarly, the limit for the duration of long-term memory has not been established, and potentially it is longer than a lifetime.

Support for the multi-store memory model

Sperling tested the existence of iconic memory (part of sensory memory). He used the so-called “partial-report technique”.

Procedure and results -

• In the experiment, participants were presented with a tachistoscopic image of a grid of alphanumeric characters.

• The image was flashed up for only 50 milliseconds. There were two conditions.

  • In the whole-report condition, participants were given an empty grid and required to fill it out with all the alphanumeric characters in the appropriate positions. They were asked to guess when they were not certain. Participants were able to recall an average of 4 out of 12 characters (35%).

  • In the partial-report condition, participants were presented with the stimulus as before but were only required to recall one of the rows from the grid. The instruction indicating which row to recall was given in the form of a sound.

  • Participants were instructed to recall the top row on hearing a high tone, the middle row on hearing a middle tone, and the bottom row on hearing a low tone.

  • Participants were usually able to recall three or four characters from the row.

  • As the row was selected at random, and after the presentation of the stimulus, we can conclude that 75–100% of the entire grid was accessible to the participant for a brief amount of time after the presentation of the stimulus. The interpretation of the findings, in line with the multi-store memory model, is that after we have been exposed to a visual stimulus, its trace stays in our memory for a short period of time. If attended to, some parts of this trace can be further consolidated and transferred into the STM.

Glanzer and Cunitz (1966) are famous for their research on the serial position effect, which serves as a support for STM and LTM being separate memory stores. The serial position effect is the tendency to recall the first and the last items on a list better than the items in the middle.

Procedure and results

• Participants were required to memorize lists of words followed by a free-recall task (There were two conditions.

  • In the first condition, participants were presented with recordings of 20-word lists consisting of common one-syllable nouns. Immediately after hearing the words, they were required to do a free-recall task for two minutes. Results of these trials clearly demonstrated the serial position effect in both its aspects: participants were better at remembering words at the start of the list (primacy effect) and at the end of the list (recency effect).

  • In the second condition, researchers introduced a delay between the end of the list and the start of recall. During the delay, participants engaged in a filer task: counting backward from a given number for 30 seconds. The filler task was meant to prevent rehearsal. The resulting data indicated that participants were still successful at recalling the words from the start of the list (primacy effect preserved), but were no longer able to recall the words from the end of the list (recency effect disappeared).

Conclusion

Glanzer and Cunitz explained that when people are hearing a list of words with the intention to memorize them, they tend to repeat the words to themselves. The first words on the list get repeated (rehearsed) more often and enter the long-term memory, which is unaffected by the delay and the filler task. However, the last words on the list are not rehearsed enough. Without rehearsal, their trace in short-term memory decays in just 30 seconds, so the recency effect disappears after the filler task. Since one of the effects disappears and the other does not, it supports the idea that STM and LTM have separate memory mechanisms behind them.

Criticisms of the Multi-Store Memory Model

• A limitation of this model is that it focuses on structure rather than the process. Even the definition of memory as a cognitive process implies that understanding how information flows is more important than seeing how many separate stores it goes through. Structure in this model is emphasized.

• Due to this lack of attention to memory as a process, the only mechanism that enables the transfer of information from STM to LTM in the original multi-store memory model is rote rehearsal. This seems to be an oversimplification that ignores various strategies that may enhance memorization

Working Memory Model

• Baddeley and Hitch (1974) developed the working memory model. This model focuses on the structure of STM.

• The working memory consists of a central executive that coordinates two subsystems: the visuospatial sketchpad and the phonological loop.

  • The visuospatial sketchpad (“the inner eye”) holds visual and spatial information.

  • The phonological loop holds sound information and is further subdivided into the phonological store (“the inner ear”) and the articulatory rehearsal component (“the inner voice”). The inner ear holds sound in a passive manner, for example, it holds someone’s speech as we hear it.

  • The inner voice, on the other hand, performs the following important functions.

    • First, it turns visual stimuli into sounds. For example, if we are shown a list of written words, we may subvocally pronounce these words, changing the modality from visual to auditory, and the words will enter our STM through the auditory channel.

    • Second, it allows the rehearsal of information held in the inner ear. By constantly repeating the words, we are increasing the duration of working memory and increasing the chances of transferring the information further into long-term memory storage.

  • The central executive is a system that allocates resources between the visuospatial sketchpad and the phonological loop. In this sense, it is the “manager” for the other two systems.

  • In 2000 Baddeley and Hitch also added the fourth component, the episodic buffer, as a component that integrates information from the other components and also links this information to long-term memory structures.

Support for the Working Memory Model

Conrad and Hull demonstrated the phonological similarity effect.

• Procedure - In their study participants were required to recall lists of letters. Some lists of letters were phonologically similar (for example, B, D, C, G, P) while others were not (for example, F, H, P, R, X).

• Results and conclusion - They found that rhyming lists were more difcult to remember. This is because the traces of similarly sounding letters (if they are encoded acoustically) are easier to confuse with each other. This supported the idea that memory for speech material uses a sound-based storage system, which we now know as the phonological store.

Baddeley, Lewis, and Vallar (1984) explored the effects of articulatory suppression on the phonological similarity effect.

• Articulatory suppression is a method of blocking the “inner voice” (articulatory rehearsal component).

• In an experimental situation, articulatory suppression is simply asking your participants to repeat a sequence of sounds (for example, the-the-the or one-two-three-one-two-three) over and over again while at the same time performing the experimental task. In doing this, the capacity of the “inner voice” is filled up. First, visual inputs cannot be recorded into sounds and hence cannot enter the phonological store. Second, auditory inputs can enter the phonological store, but their rehearsal will be impossible.

• Results - The results indicated that there was a phonological similarity effect between conditions 1 and 2 (that is, rhyming words were significantly harder to recall than non-rhyming words), but there was no phonological similarity effect between conditions 3 and 4 (for written material, rate of recall for both rhyming and non-rhyming words was the same).

  • When articulatory rehearsal is inhibited, spoken information can still enter the phonological store (the “inner ear”) directly.

  • Since rhyming words sound similar and create similar traces, these traces are easier to confuse so we observe the phonological similarity effect.

  • Written information can also enter working memory, but it does not get recorded into sounds. Presumably, it enters the visuospatial sketchpad. As the information is coded visually, the traces are not that easy to confuse, and the phonological similarity effect is not observed.

For the central executive, one of the studies that provided supporting evidence was conducted by Alan Baddeley (1996).

• The starting point of his reasoning was that since the theoretical function of the central executive is to distribute and switch attention, it should be inhibited by tasks that require attentional switches, and at the same time, it should not be inhibited by tasks that do not require attentional switches.

• Procedure - In the study, participants were required to produce random sequences of digits by pressing keyboard keys at the rate of one per second, determined by a metronome. To produce a random sequence of digits, you have to use your attentional resources because you need to take into account the previous digits that you have selected. So the dependent variable in this experiment was the randomness of the digit sequence (participants were required to produce sequences of 100 digits overall). The more random the sequence, the better the central executive performed at controlling this cognitive task. Simultaneously, participants were required to engage in one of three tasks, at the same rate of one unit per second.

  1. Recite the alphabet (A, B, C, and so on).

  2. Count (1, 2, 3, and so on).

  3. Alternate between letters and numbers (A, 1, B, 2, C, 3, and so on).

• Results - results showed that “whereas neither counting nor reciting the alphabet had a detectable effect on the randomness of key pressing, the concurrent alternation task markedly reduced randomness”. So it was concluded that this constant switching of retrieval plans is performed by a separate memory system (the central executive).

  Evaluation of the working memory model

Strengths

• More sophisticated than the multi-store memory model.

• The model can integrate a large number of findings from work on short-term memory.

• The working memory model does not overemphasize the role of rehearsal.

Weaknesses

• Models of this degree of complexity are harder to test empirically.

• The model is difcult to falsify.

• The working memory only involves STM and does not take into account other memory structures, such as LTM and sensory memory.

Schema Theory

Concepts of schema theory

• Cognitive schemas - mental representations that organize our knowledge, beliefs, and expectations.

• Mental representations guide behavior.

• Mental representations need to be quite stable, deeply rooted, and organized.

• Schemas are derived from prior experience.

• schemas influence memory processes at all stages—encoding, storage, and retrieval.

Schemas influence encoding

• Bransford and Johnson (1972) carried out an experiment in which they investigated the effect of context on comprehension and memory of text passages. It followed an independent measures design.

• Participants were instructed, after hearing the passage, to recall it as accurately as they could, and to write down as many ideas as possible. They were given seven minutes to do that. The five conditions were:.

  1. No context (1): participants simply heard the passage.

  2. No context (2): participants heard the passage twice.

  3. Context before: prior to hearing the passage participants were provided with a context picture and given 30 seconds to study it.

  4. Context after: the same picture was shown, but after participants heard the passage.

  5. Partial context: a context picture was provided before the passage. This picture contained all the objects, but the objects were rearranged.

• Results - the “context before” condition made a difference. Hearing the same passage twice makes almost no difference in terms of comprehension and memory; presenting the context after the passage is no good, and a partial context (which shows the objects but not the relations between them) is only marginally better than no context at all.

Conclusion - After the schema has been created (or activated), it influences the organization of our knowledge. Arguably, the idea units from the text passage are more effectively encoded because, in the process of encoding, they are linked with the schema.

Schemas influence retrieval

Anderson and Pichert (1978) demonstrated that schemas not only influence the memory process of encoding (transferring information to LTM) but also have an influence at the stage of retrieval (retrieving information from LTM).

1. A series of tasks were performed by the participants.

2. They are assigned either the homebuyer or the burglar perspective.

3. They were asked to read a text passage about what two boys did at the home of one of the boys when they were skipping school. The passage contains a number of points of interest to a burglar or real estate agent.

4. They are given two minutes to read the passage.

5. There is a filler verbal task for 12 minutes.

6. . Following this, they were given two blank pages and asked to reproduce the story in as much detail as possible.

7. There is another five-minute filler task.

8. Next, the instructions require you either to keep the same initial perspective (homebuyer or burglar) or to change it.

9. You are required to recall the text for a second time (without reading it again). In all there were four groups of participants.

Results revealed the following pattern.

• For the first recall, the group that had the burglar perspective recalled more burglar information whereas the group that had the homebuyer perspective recalled more homebuyer information.

• The people who changed perspective recalled more information important to the second perspective but unimportant to the first.

• Subjects who changed perspective (groups 3 and 4) recalled an additional 7.1% of the now-important information.

• On the other hand, subjects who did not change perspective (groups 1 and 2) recalled 2.9% less of the still unimportant information.

  Conclusion - Schemas organize knowledge that is stored in our memory. Any new information that we acquire does not just get passively registered; it is actively perceived through existing schemas. They start acting at the stage of encoding.

Types of schemas

• social schemas—mental representations about various groups of people, for example, a stereotype

• scripts—schemas about sequences of events, for example, going to a restaurant or making coffee

• self-schemas—mental representations about ourselves.

Bottom-up and top-down processing

• Bottom-up information processing occurs when the cognitive process is data-driven; perception is not biased by prior knowledge or expectations.

• Top-down processing occurs when your prior knowledge or expectations (schemas) act as a lens for the information that you receive and process.

• Top-down processing is considered a bad thing because it can potentially lead to a variety of biases. However, it is actually very necessary.

• Using schematic processing, we see patterns in otherwise unstructured stimuli (pattern recognition) and find meaning in those patterns (effort after meaning).

• Evolution has predisposed us to see certain patterns even when the stimulus is not clear or is vague because these patterns might be potentially important to us.

Thinking and decision making

• Decision-making is a cognitive process that involves selecting one of the possible beliefs or actions, that is, making a choice between some alternatives. It is closely linked to thinking.

• Thinking and decision-making are complex higher-order cognitive processes.

Normative Models

• Normative models describe the way that thinking should be.

• They assume that unlimited time and resources are available to make a decision. They define what is right and wrong, correct and incorrect, and effective and ineffective.

• One example of a normative model of thinking is formal logic.

• Another example of a normative economic model is the theory of probability.

• Utility theory is the normative model for decisions involving uncertainty and trade-offs between alternatives. According to this theory, the rational decision-maker should calculate the expected utility for each option and then choose the option that maximizes this utility.

Descriptive Models

• Descriptive models show what people actually do when they think and make decisions.

• They focus on an accurate description of real-life thinking patterns and the main measure of effectiveness for such models is how closely the model fits observed data from various samples of participants.

The theory of reasoned action and planned behavior

• The theory of reasoned action (TRA) aims to explain the relationship between attitudes and behaviors when making choices.

• Was proposed by Martin Fishbein in 1967.

• An individual’s choice of a particular behavior is based on the expected outcomes of that behavior.

• There are two factors that determine behavioral intention: attitudes and subjective norms.

• In 1985 the theory was extended and became what is known as the theory of planned behavior (TPB). This theory introduced the third factor that influences behavioral intentions: perceived behavioral control.

  Study in focus - Albarracin et al (2001) conducted a meta-analysis of TRA and TPB as models of condom use. The meta-analysis comprised 42 published and unpublished articles and a total of 96 data sets.

  Conclusion - people are more likely to use condoms if they have previously formed the corresponding intentions. These intentions to use condoms appear to derive from attitudes, subjective norms, and perceived behavioral control.

Reliability of cognitive processes

Unreliability of Memory

• Schemas can determine what you do and do not remember even after the information has been coded and stored in long-term memory.

• Depending on the schema you are using, you will nd it easier to recall some details.

• Retrieval of information from LTM may depend on whether or not you are using a particular schema.

The theory of reconstructive memory and eyewitness testimony

• Memory, rather than being the passive retrieval of information from the long-term storage, is an active process that involves the reconstruction of information, the theory of reconstructive memory.

• Loftus and Palmer suggest that memory for some complex event is based on two kinds of information: information obtained during the perception of the event and external post-event information.

• Over time, information from these two sources is integrated in such a way that we are unable to tell them apart.

• Applied to the study, this means that subjects who were given the question with the verb “smashed into” used this verb as post-event information suggesting that the accident had been severe.

• This post-event information was integrated into their memory of the original event, and since broken glass is commensurate with a severe accident, these subjects were more likely to think that they had seen broken glass in the film.

These findings can also be interpreted from the perspective of schema theory: the high-intensity verb “smashed” used in the leading question activates a schema for severe car accidents. Memory is then reconstructed through the lens of this schema.

Misleading information, recognition and visual memory

• Loftus, Miller and Burns (1978) carried out an experiment with the aim of investigating how verbal information supplied after an event influences a witness’s visual memory for that event.

• Procedure - In the recognition procedure, participants were shown a series of 30 color slides depicting successive stages in a car–pedestrian accident. T

  • The slides featured a red Datsun travelling along a side street toward an intersection with either a stop sign (for half the subjects) or a yield sign (for the other half).

  • In the remaining slides, the Datsun turned right and knocked down a pedestrian who was crossing at the crosswalk.

  • After viewing the slides, the subjects answered a series of 20 questions. Question 17 on the list was either “Did another car pass the red Datsun while it was stopped at the stop sign?” for half the subjects or the same question with the words “stop sign” replaced by “yield sign” for the other half. This resulted in a 2-by-2 experimental design.

• Results - that subjects who received misleading post-event information were able to correctly recognize the slide actually seen in 41% of the cases, whereas subjects who received consistent post-event information made a correct choice in 75% of the cases.

Biases in thinking and decision making

System 1 and 2 thinking

• Daniel Kahneman in 2003 proposed anextension to the information-processing approach by differentiating between two independent systems, system 1 and system 2.

• According to the theory, system 1 thinking is fast, instinctive, emotional, automatic and relatively unconscious, whereas system 2 thinking is slower, more analytical, logical, rule-based and conscious.

Emotion and cognition

The influence of emotion on cognition

• Cognitive appraisal can be a mediator between physiological arousal and the subsequent emotional response.

The interaction between cognition and emotion in behaviour is bidirectional.

The theory of flashbulb memory

• The theory of flashbulb memory was proposed by Roger Brown and James Kulik in 1977.

• Flashbulb memories are vivid memories of the circumstances in which one first learned of a surprising and emotionally arousing event.

• There are two variables that had to attain sufciently high levels in order for flashbulb memory to occur: surprise and a high level of personal consequentiality (which causes emotional arousal).

• If these variables reach a sufficient level, this triggers a maintenance mechanism: overt and covert rehearsal which reinforces the degree of elaboration of the event in memory.

• The mechanism of formation is a photographic representation of events that are surprising and personally consequential and therefore emotionally arousing.

• The mechanism of maintenance includes overt rehearsal (conversations with other people in which the event is reconstructed) and covert rehearsal (replaying the event in one’s memory).

Cognitive processing in the digital world

Digital technology and cognitive skills

• Interaction with digital technology has the potential to improve cognitive processing skills.

• Studies have demonstrated positive effects of videogame play on experimental tasks involving:

  • hand–eye coordination

  • reaction time

  • spatial visualization

  • mental rotation

Sanchez (2012) aimed to study whether there were positive effects of video games on science learning. It had been established already that some visuospatial games improve performance on visuospatial ability tests.

Procedure - University students were randomly divided into two groups: the spatial training group who played a first-person shooter and the non-spatial training group who played a game that involves making words. After playing the game, participants read a complex text about plate tectonics. After reading, participants were required to apply the learned concepts in a novel situation by writing an essay entitled “What caused Mt St Helens to erupt?”

Results - Participants from the spatial training group (the ones that played Halo) scored better on the essay. It was concluded that spatial training with the use of video games potentially can improve understanding of spatial relations.

One study looked at the effects of training with a computer simulator on golf-playing skills.

Fery and Ponserre (2001) studied men with no prior golf experience.

Procedure - One group of participants engaged with the simulation with the intent to improve golf putting (the learning group), one group played the simulation to simply enjoy the game (the entertainment group), and one was a control group.

Results - golf putting improved in both the entertainment and the learning group; however, improvement in the learning group was most signicant.

It was concluded that the usefulness of video game simulations on actual golf-putting skills depended on two conditions, credibility and motivation.