π Lecture 07 β ATTENTION
π Overview: This lecture introduces attention as a fundamental cognitive process, exploring its limited-capacity nature through various metaphors and experimental evidence. It examines how attention selects information from sensory memory, the limitations of this selection process, and classic experimental paradigms like dichotic listening that reveal how attention operates, including debates about whether selection occurs early or late in processing.
ποΈ Topics Covered
The lecture covers the conceptualization of attention as a limited mental resource using multiple metaphors (spotlight, filter, bottleneck, energy, workspace, demons). It discusses the single-minded nature of attention and its capacity limitations, explores how these limitations manifest in sensory tasks like whole report performance, presents the dichotic listening paradigm by Cherry and Moray, introduces the Early Selection Model of attention, and examines research by Gray and Wedderburn showing that attention can track meaning across ears.
π Lecture Summary
What is attention?
Attention is conceived of as being a very limited mental resource. It plays a crucial role in the stage of selection of information from sensory memory, where most information is discarded and only some is selected for further processing. Numerous metaphors help us think about the limited-resource characteristics of attention.
Common metaphors for attention include:
- Spotlight/beam β like a stage spotlight that falls on one person, attention focuses on one input
- Filter/sieve β like listening to one conversation at a party while filtering out others
- Bottleneck/narrow lane β like traffic that can only flow through a narrow passage
- Limited resource β like skilled manpower or petrol, only enough for limited tasks
Additional limited-resource metaphors:
- Energy: Like electricity with a fixed current, attention has a fixed energy supply allocable to only so many tasks. If allocated to more, performance degrades or a "fuse would blow."
- Spatial or workspace: Like a small room that can only fit a few employees, limited workspace means only so many tasks can be performed.
- Demons: Attention as a small set of agents (demons) that can perform tasks but only one at a time.
π Definition β Attention: a very limited mental resource that selects information from sensory memory for further processing.
Single-mindedness
Attention is single-minded in sense, not double-minded. This single-mindedness means that only enough energy, only enough workspace, or only a single attention demon is available for one task or process. So it means one thing at a time. Attention has a capacity to perform only one demanding task. Attention cannot perform two demanding tasks simultaneously.
But what about walking and talking? Driving and talking, and smoking? Tasks that are practiced to the point at which they do not make excessive demands can be performed simultaneously. We cannot simultaneously do mental addition and carry on a conversation because each activity involves multiple attention-demanding subcomponents. Whether or not attention is truly single-minded, its capacity is severely limited.
π Example: Walking and talking can be done together because walking is so practiced it makes few demands, but mental addition and conversation cannot be done simultaneously because both involve multiple attention-demanding subcomponents.
π‘ Why this matters: Understanding the single-minded nature of attention explains why multitasking often fails and why practice can make some tasks automatic, freeing up attentional resources.
Limitations in sensory tasks
The limited capacity of attention is the root cause of the reporting limitations demonstrated in visual and auditory reporting tasks. Attention appears to be the real reason for whole report performance. Limitations of the icon and echo are actually the limitations of attention. All the information gets into sensory memory, but to be retained, each unit of information must be attended to and transformed into some more permanent form. Sensory information needs to be stored in a form that can be retained. Selection for this transformation happens only to attended items. Most information is lost because attention has limited capacity.
Dichotic Listening Tasks
Cherry (1953) and Moray (1959) conducted experiments on how subjects select what sensory input they attend to, using a dichotic listening task.
In a dichotic listening experiment, subjects wear a set of headphones. They hear two messages β one ear presented with one message, the other ear with another message. Subjects pay attention to (shadow) one message and tune out the other. Psychologists discovered that very little about the unattended message is processed in a shadowing task. Subjects cannot tell what language was spoken or report any of the words spoken even if the same word was repeated over and over again. Subjects reported hearing very little from the other ear.
π Example (Shadowing Paradigm):
- Messages in Left Ear: ran, house, Ox, Cat
- Messages in Right Ear: Tea, job, books, look
- People asked to attend to left ear report: ran, house, Ox, Cat
- They report nothing from the unattended ear
Early Selection Model
Early selection means information is selected early, soon after the sensory store. It happens early on in this system. Higher level processing is not done. The meanings of the words do not have an impact at this stage in this model. Attention itself is a lower level process.
The Early Selection Model shows information coming from right and left ears, but before going to bottom line, information from one ear blocks the other ear's information. According to this model, attention is an early selection process and a low-level process.
π Definition β Early Selection Model: a model proposing that attention selects information for processing soon after the sensory store, before higher-level meaning analysis occurs.
Attention and meaning
Gray and Wedderburn (1960) conducted an experiment at Oxford demonstrating that subjects were quite successful in following a message that jumped back and forth between ears. This challenged the early selection view.
π Example (Gray & Wedderburn's experiment):
- Left Ear: John β Eleven β books
- Right Ear: Eight β writes β Twenty
- Instructed to shadow the meaningful message, subjects reported: John writes books
Thus, subjects are capable of shadowing a message on the basis of meaning rather than physical ear. This suggests that some semantic processing of unattended information may occur before selection.
β Key Takeaways
Attention is a severely limited mental resource that selects information from sensory memory for further processing, as illustrated by multiple metaphors including spotlight, filter, bottleneck, energy, workspace, and demons. While attention is single-minded and cannot perform two demanding tasks simultaneously, practiced tasks can become automatic and require fewer attentional resources. The limited capacity of attention explains why most sensory information is lost β each unit must be attended to for retention. Dichotic listening experiments (Cherry, Moray) show that very little of unattended messages is processed, supporting early selection views. However, Gray and Wedderburn's findings that subjects can shadow messages based on meaning across ears suggest that at least some semantic processing may occur before selection, challenging simple early selection models.
π§ Quick Revision Questions
- What are the five main metaphors used to describe attention's limited-resource characteristics, and what does each emphasize about attention?
- Why can people walk and talk simultaneously but cannot do mental addition and carry on a conversation at the same time?
- In Cherry and Moray's dichotic listening experiment, what information were subjects able to report from the unattended ear?
- According to the Early Selection Model, at what point in processing does attention select information, and what level of processing occurs before selection?
- How did Gray and Wedderburn's 1960 experiment challenge the Early Selection Model of attention?
π Lecture 8 β Attention and meaning
π Overview: This lecture continues the discussion of attention by examining whether attention operates as an early or late selection process. Groundbreaking experiments by Gray & Wedderburn and Treisman demonstrated that meaning, not just physical characteristics, can guide selective attention, leading to the development of competing theoretical models.
ποΈ Topics Covered
This lecture covers the Gray & Wedderburn experiment showing attention can use meaning, Treisman's experiment on early versus late selection, three major attention models including Treisman's attenuator model, Broadbent's filter model, and Norman's late selection model, and concludes with a comparison of early and late selection theories.
π Lecture Summary
ATTENTION (continued)
Two undergraduates at Oxford, Gray and Wedderburn (1960) conducted an experiment challenging the idea that attention is a low-level early selection process. If attention selects only on physical characteristics like ear of presentation, then meaning should not matter. Subjects were given different words to each ear simultaneously: β’ Left Ear: John Eleven books β’ Right Ear: Eight writes Twenty
When instructed to shadow the meaningful message, subjects reported: "John writes books." This demonstrated that subjects are capable of shadowing a message based on meaning rather than physical ear, switching between ears to follow the meaningful content.
Implications
The Gray and Wedderburn experiment proved that attention can use meaning, which is a higher-level process. Subjects switched some information and selected meaningful words regardless of whether they were presented in the left ear or right ear. This suggests that attention is probably not an early selector but a late selector that operates after meaning has been processed.
Triesman's experiment
Treisman (1960) conducted an experiment to determine whether attention is an early selector or late selector. Subjects were asked to shadow the message in the left ear. The messages were: β’ In Left ear: I was going there when/ China, smoke, lovely, chirping β’ In Right ear: books, chairs, tables, elephant/ I saw a bright flash
Many subjects switched ears to follow the meaningful message, replicating the finding that meaning guides attention.
Attention models
1. Treisman's Model
Treisman proposed an early selection model based on her experiment. In this model, there is no filter but an attenuator. The attenuator weakens unattended signals rather than completely blocking them. After the attenuator, information is processed in a mental dictionary where some items are more important than others (e.g., our name is more important than "table" or "fire"). This model explains selection on the basis of meaningful information.
π Definition β Attenuator: A device in Treisman's model that weakens unattended signals instead of completely filtering them out. π‘ Why this matters: The attenuator explains how we can still notice important information (like our name) from unattended channels, whereas a strict filter would block it completely.
2. Broadbent's experiment
Broadbent's model shows that senses take in information within their limited capacity. This information goes into short-term store (sensory store), which has unlimited capacity but retains information for only a very brief period. Then a selective filter tunes out extra information and passes selected information through a limited capacity channel to other systems. One path goes back to short-term store as a feedback loop. The limited-capacity channel also passes information to the store of conditional probabilities of past events, which then moves again towards the selective filter. Information then goes to a system for varying output until some input is secured, and then moves towards effectors (physical responses to attention).
π Definition β Selective Filter: In Broadbent's model, a mechanism located soon after the short-term store that selects information based on physical characteristics, not meaning. π Key feature: Broadbent's model does NOT select on the basis of meanings β selection occurs early, before semantic processing. π‘ Why this matters: This is the original "early selection" model and served as the foundation for all subsequent attention theories.
3. Norman's Model
Norman proposed a late selection model. He said we get all information that is stored in short-term memory, then in long-term memory. In long-term memory, we check its relevance; when relevance is determined, we select information and then pay attention. This model is opposite to early selection models because all information is processed for meaning before selection occurs.
π Definition β Late Selection Model: A model where all incoming information is processed for meaning before attention selects what to focus on.
There are two most important models:
1. Early selection model β comes in two types:
- Filter model (Broadbent): Some things pass, some are stopped completely
- Attenuator model (Treisman): Some information is weakened, some is strengthened
2. Late selection model
In the attenuator model, the mental dictionary contains items that are very important to us. These important items are strengthened by the attenuator, allowing them to capture attention even from unattended channels.
π Definition β Filter vs. Attenuator: A filter completely blocks unattended information; an attenuator weakens it so important information can still be noticed. π Key distinction: Early selection occurs before meaning analysis; late selection occurs after meaning analysis.
β Key Takeaways
The most critical thing to remember is that attention can operate based on meaning, not just physical characteristics, as proven by Gray & Wedderburn's and Treisman's experiments where subjects switched ears to follow meaningful messages. The two major competing theories are early selection (Broadbent's filter model and Treisman's attenuator model) and late selection (Norman's model). Treisman's attenuator model is particularly important because it explains how meaningful information like one's own name can break through from unattended channels. Broadbent's original filter model places selection early, before semantic processing, while Norman places it after full semantic analysis in long-term memory.
π§ Quick Revision Questions
- What did Gray and Wedderburn's "John Eleven books / Eight writes Twenty" experiment demonstrate about attention?
- What is the key difference between Broadbent's filter model and Treisman's attenuator model?
- In Treisman's model, what is the role of the "mental dictionary" in selective attention?
- What is the fundamental difference between early selection and late selection models of attention?
- According to Norman's model, where does the selection of information for attention occur?
π Lecture 09 β Capacity Models
π Overview: This lecture examines attention through the lens of capacity models, shifting from earlier bottleneck theories to more flexible frameworks. It explores Kahneman's capacity model, which proposes a general limit on mental work capacity that can be flexibly allocated, and introduces the multimode theory that integrates capacity with stage of selection. Understanding these models is crucial for explaining how people manage multiple tasks and why interference occurs.
ποΈ Topics Covered
This lecture covers the shift from bottleneck to capacity models of attention, Kahneman's capacity model including arousal, enduring disposition, and momentary intentions as allocation policies, the World Trade Centre example illustrating attention allocation failure, comparison between bottleneck and capacity theories, the interaction between capacity and stage of selection, and the multimode theory which proposes that selection can occur at either early or late stages depending on task demands and intentions.
π Lecture Summary
Attention (continued)
Psychologists have become more interested in the capacity demands of different tasks (Kahneman, 1973). Different tasks require different amounts of attention, and attention can be diverted from one task to another. Tasks require mental effort. People may have some control over where the bottleneck occurs (Johnston & Heinz, 1978). π‘ Why this matters: This introduces the idea that attention is not fixed but can be flexibly managed, challenging earlier rigid filter models.
Kahnemanβs Capacity Model
"Attention and Effort" was a major work of Kahneman (1973). He shifted the focus from bottleneck to capacity. There is flexibility in attention β we can change our attention from one thing to another. Evidence shows our bottleneck is actually adjustable and can move from early to late. There is a general limit on a personβs capacity to perform mental work. A person has considerable control over how this capacity is allocated.
π Definition β Daniel Kahneman: A pioneer of cognitive psychology who has been working at Princeton University, contributing not only to attention but other areas.
Kahnemanβs Capacity Model (Diagram Explanation)
In this model, many miscellaneous determinants impact the sensory system. Something happens that triggers arousal β meaning some activity starts. Arousal has many manifestations. Available capacity of attention will be allocated depending on the state of arousal. There is an allocation policy. Then there are some possible activities β the amount of attention paid on a task. There is a feedback loop: we evaluate how much attention is needed for the task, then readjust more capacity to that task.
π Definition β Arousal: A physiological state that influences the distribution of mental capacity in the various tasks.
π Definition β Enduring Disposition: An automatic influence where people direct their attention.
π Definition β Momentary Intentions: A conscious decision to allocate attention to certain tasks or aspects of the environment.
π Example: World Trade Centre A Boeing 707 was flying in cloudy weather at an altitude 200 feet below the top of the WTC. The airport controller reported this. An alarm buzzed in the airport control tower to signal danger. The controller radioed the crew to turn around and climb up to 3000 feet. The controller was monitoring seven other planes at the time. His attention was diverted to other planes, so he could not pay attention to that plane. The alarm diverted his attention toward that plane.
Bottleneck vs. Capacity
Both models predict that simultaneous activities are likely to interfere with each other, but they attribute the interference to different causes.
- Bottleneck: According to this model, the same mechanism is required to perform 2 incompatible tasks. It is specific β it says the same mechanism is needed for those tasks.
- Capacity: Demands of 2 activities exceed available capacity, then there is a problem. It is non-specific to the task. Total demands of the task is an important variable.
Both kinds of interference occur (specific and capacity, early selection and late selection). Both kinds of theories are necessary.
Capacity & Stage of Selection
All experiments show flexibility of attention β we can divert our attention from one task to another. These also show an interaction between bottleneck and capacity theories (Johnston & Heinz, 1978). The listener has control over the location of the bottleneck. The location can vary from early mode of selection (before recognition) to late mode of selection (after semantic analysis, meaning meaning analysis). There is a need to combine capacity model and stage of selection.
Multimode Theory
A theory that proposes that peopleβs intentions and the demands of the task determine the information processing stage at which information is selected. Demands of the task require greater mental efforts. Peopleβs intentions and the demands of task decide where attention is paid. We select on the basis of meanings if there are two competing tasks.
Early Selection vs. Late Selection
According to this theory, both early and late selection can occur. Attention is flexible β we can move our bottleneck and filter. But there must be interaction between bottleneck and capacity. We can shift our bottleneck to the low level in information processing when we pay attention to the physical properties of the task. We have a capacity to switch our bottleneck from early to late or late to early. Late selection will affect the perception of primary message because more information is selected about the secondary task.
β Key Takeaways
Kahneman's capacity model revolutionized attention theory by shifting focus from a fixed bottleneck to a flexible capacity system with a general limit on mental work. The allocation of attentional capacity depends on arousal, enduring dispositions (automatic), and momentary intentions (conscious decisions). Bottleneck and capacity theories are both necessary: bottleneck explains specific interference from incompatible tasks requiring the same mechanism, while capacity explains non-specific interference when total demands exceed available capacity. The multimode theory (Johnston & Heinz, 1978) integrates these views, showing that people can control the location of the bottleneck β shifting between early selection (physical properties) and late selection (semantic analysis) depending on task demands and intentions. Late selection consumes more capacity because more information is processed about secondary tasks, which can impair primary task perception.
π§ Quick Revision Questions
- What did Kahneman (1973) shift the focus of attention research from, and what did he propose instead?
- What are the three factors in Kahneman's model that influence the allocation policy of attention?
- In the World Trade Centre example, why did the controller fail to notice the Boeing 707's dangerous altitude?
- According to the multimode theory, what determines whether selection occurs at an early or late stage?
- How does late selection affect the perception of the primary message compared to early selection?
π Lecture 10 β Multimode Theory (continued)
π Overview: This lecture continues the discussion of Multimode Theory, which proposes that peopleβs intentions and task demands determine the stage at which information is selected for processing. It presents a key experiment testing this theory using selective listening with a subsidiary reaction-time task, and examines how capacity demands change based on whether selection occurs early (physical cues like pitch) or late (semantic cues like meaning).
ποΈ Topics Covered
This lecture covers the continuation of Multimode Theory, the main assumption of Capacity Theory, the concept of reaction time and accuracy, a detailed selective listening experiment with conditions of no list, one list, and two lists (using pitch or meaning), the predictions of Multimode Theory, results for both subsidiary and primary tasks, and the implications and summary of findings. The experiment demonstrates that selective attention requires capacity, that more capacity is needed for late selection (meaning) than early selection (pitch), and that reaction times and error rates increase accordingly.
π Lecture Summary
Multimode Theory (continued)
A theory which proposes that peopleβs intentions and the demands of the tasks determine the information processing stage on which information is selected. This means that selection can occur early, based on physical features, or late, based on semantic meaning, depending on what the task requires.
Experiments to test multimode theory A series of 5 experiments was conducted to measure the amount of capacity required to perform a task and to record how quickly a person responds to a subsidiary task (reaction time). The main task was a selective listening task, where different voices are presented in both ears, and subjects are asked to attend to only one voice or list and repeat it. The subsidiary task was to respond to a randomly appearing light signal by pressing a button.
π Definition β Subsidiary Task: A task that typically measures how quickly people can react to a target stimulus in order to evaluate the capacity demands of the primary task.
Every theory has assumptions; if an experiment fulfills those assumptions, then the theory is supported. Every theory predicts some phenomenon, and these predictions are tested through experiments. If predictions are met, the theory is accepted; otherwise, it is rejected.
Capacity Theory
The main assumption of Capacity Theory is: The greater the portion of capacity allocated to selective listening (the primary task), the less should be available for monitoring the signal light, causing longer reaction times. In other words, if the selective listening task requires more capacity, then the subsidiary task's capacity to monitor the light signal is reduced, leading to a longer reaction time.
π‘ Why this matters: This assumption directly links attention to a limited pool of cognitive resources, implying that tasks compete for a finite capacity.
Reaction Time & Accuracy
The time it takes for a subject to respond to a stimulus is called reaction time. The longer the time taken, the more difficult the task. If the task is demanding, subjects are likely to make mistakes and take a long time. With a more demanding task, there are more mistakes and less accuracy. A less demanding task leads to more accuracy. Sometimes reaction time is slow but accuracy is still present.
Experiment
In this experiment, pairs of words were presented simultaneously to both ears through headphones. Undergraduates at the University of Utah were asked to attend to different voices. The stimuli were selected either by the pitch of the voice (male/female) or by semantic category (cities, occupations).
The conditions of the experiment were:
- No lists: No list of stimuli was presented in either ear.
- One list: Only one list was presented in one ear (e.g., a male or female voice).
- Two lists: Two different lists of stimuli were presented in both ears. In this condition, subjects had to first understand meanings (late selection). In one ear, a male or female voice was presented; in the other ear, different city names or occupation names were presented.
The rationale was: those using pitch (male/female) were using physical information and could use an early, sensory mode of selection because the two messages were physically different. Those using meaning (cities, occupations) had to use a late, semantic mode of selection because it was necessary to know the meaning of the word before categorizing it.
Predictions of Multimode theory
Multimode theory predicts that more capacity is required to perform at a late mode of selection. Therefore, use of the semantic mode would cause slower reaction times to the light signal and more errors on the selective listening task. Late selection requires more capacity. In the no list condition, performance is best; in the one list condition, performance is better; in the two lists condition, performance is worst.
Results of subsidiary tasks
Performance on Subsidiary Task
- When No list was given, the reaction time was 310 ms
- When there was one List (either female voice or cities names), the reaction time was 370 ms
- In Two Lists condition (pitch), the reaction time was 433 ms
- In Two Lists condition (meaning), the reaction time was 482 ms
Results of Primary Task
Percentages of Errors
- In case of one List, the error percentage was 1.4%
- In case of Two Lists (Pitch), the error percentage was 5.3%
- In case of Two Lists (Meaning), the error percentage was 20.5%
Implications
The implications of this experiment are:
- Selective Attention requires capacity β it is not cost-free.
- Reaction Time is slower for two lists condition over one β more information to process increases difficulty.
- Amount of capacity required increases from early to late selection β semantic processing consumes more resources than physical feature processing.
- Reaction Time is slower for meaning than for pitch β categorizing by meaning is more demanding than differentiating by voice.
Summary
- Attention is flexible β people have the choice of how best to use it. People can allocate attention to a task according to their will.
- Task difficulty decreases with practice β over time, tasks become less demanding and require less capacity.
β Key Takeaways
This lecture demonstrates that attention is not a fixed filter but a flexible, capacity-limited system. The core finding is that the demands of the primary task determine how much capacity remains for secondary tasks, with late-selection (semantic) tasks requiring more capacity than early-selection (physical) tasks. The experiment clearly shows that reaction times and error rates increase systematically as the complexity of the selective listening task increases, from no list (fastest, fewest errors) to two lists with meaning (slowest, most errors). Students must remember that Capacity Theory assumes a finite pool of attentional resources, and that Multimode Theory predicts that late selection consumes more capacity than early selection. The specific numerical results (310 ms, 370 ms, 433 ms, 482 ms for reaction times; 1.4%, 5.3%, 20.5% for errors) are essential for understanding the pattern of evidence supporting these theories.
π§ Quick Revision Questions
- What is the main assumption of Capacity Theory regarding the relationship between the primary task and the subsidiary task?
- In the experiment described, what were the two different modes of selection (early vs. late) and what cues did they rely on?
- What were the reaction times for the subsidiary task under the "no list," "one list," "two lists (pitch)," and "two lists (meaning)" conditions?
- Why does Multimode Theory predict that the semantic (meaning) mode requires more capacity than the pitch mode?
- What were the error percentages for the primary task under the "one list," "two lists (pitch)," and "two lists (meaning)" conditions, and what does this suggest about task difficulty?
π Lecture 11 β RECAP OF LAST LESSONS
π Overview: This lecture revisits bottleneck and capacity models of attention, then introduces automaticityβhow well-practiced skills require minimal mental effort. It examines the pros and cons of automatic processing, criteria for automaticity, and key experiments (Schneider & Shiffrin) that contrast automatic versus controlled processing using visual scanning tasks.
ποΈ Topics Covered
The lecture begins with a recap of filter and capacity models of attention. It then defines automatic processing, its advantages and disadvantages, and criteria for automaticity (Posner & Snyder, 1975). It presents Shiffrin & Schneiderβs perspective that automaticity is a matter of degree. The core experimental section describes Schneider & Shiffrinβs visual array scanning studies, contrasting same-category vs. different-category conditions, and discusses implications for attentional load and capacity.
π Lecture Summary
RECAP OF LAST LESSONS
Attention can be explained by bottleneck or filter models as well as by capacity models. Filter models assume that interference in attention is caused by a filter or an attenuator. Some filter models assume that filters occur early, before recognition. Others assume that filter occurs late, after semantic analysis. Capacity models assume that interference is caused by overload on the capacity.
AUTOMATICITY
Automatic Processing Tasks vary considerably in the amount of mental effort required to perform them. Some skills become so well-practiced and routine that they require very minimal mental capacity. Cognitive psychologists use the term automatic processing to refer to such skills. The more a process has been practiced, the less attention it requires, and there is speculation that highly practiced processes require no attention at all β such processes are referred to as automatic.
Pros and cons
- It allows us to perform routine activities without much concentration or mental effort; it does not require much attention.
- Automatic processes complete themselves without conscious control by the subject.
- We may make silly mistakes.
- We may fail to remember what we did.
- We are not able to show others how we do a task.
When is a skill automatic? According to Posner and Snyder (1975), a skill is automatic if it:
- Occurs without intention
- Does not give rise to conscious awareness
- Does not interfere with other mental activities
Automaticity: Another Perspective Shiffrin & Schneider (1977) argue that it is best to think of automaticity as a matter of degree rather than a distinct category. A nice demonstration of the way practice affects attentional limitations is the study reported by Underwood (1974) on the psychologist Moray, who has spent many years studying shadowing (split attention studies). Moray can report most of the unattended channels; for him, shadowing has become automatic. Through a great deal of practice, the process of shadowing has become partially automated.
More experiments Schneider & Fisk (1982), Schneider & Shiffrin (1977), Shiffrin & Dumais (1982), and Shiffrin & Schneider (1977) performed a series of experiments contrasting controlled vs. automatic processing.
They said automatic processes complete themselves without conscious control by the subject. In the visual and auditory report tasks reviewed earlier, the registering of the stimuli in sensory memory is an automatic process. Many aspects of driving a car and comprehending language appear to be automatic. Controlled processing seems to require conscious control. Many higher cognitive processes, such as performing mental arithmetic, are controlled.
In their experiments, subjects were required to scan visual arrays. The subjects were given a target letter or number and instructed to scan a series of visual displays for the target.
Visual arrays
1st condition:
- J (letter to be recognized)
- GK
- MF
2nd condition:
- 8 (number to be recognized)
- MN
- L 8
The task Two factors are varied:
- Frame size: Each frame has one, two, or four characters on it.
- Relationship between the target item and the items on the frames: In the same category condition, the target is a letter and all the characters on the frame are letters. In the different category condition, the target is a number surrounded by letters.
If the target appears on the frame, subject responds "yes"; if it doesn't appear on the frame, the subject responds "no".
Results
- In the different category condition: Reaction Time was 80 milliseconds, and Accuracy was 95%.
- In the same category condition: Reaction Time was 400 milliseconds, and Accuracy was also 95%.
In the different category condition: No effect of frame size.
In the same category condition: Accuracy and RT deteriorated as frame size increased.
Schneider and Shiffrin argued that before coming into the laboratory, subjects were so well practiced at detecting a number among letters that this process was automatic. In contrast, when subjects had to identify a letter among letters, controlled processing was needed. In this situation, subjects had to attend separately to each letter in the frame and compare it with the target. All these steps took time, and thus subjects were able to inspect each frame properly and achieve respectable levels of performance only when slides were presented slowly. Also, the more letters that were in a frame, the more slowly the frames had to be presented, since subjects had to check each letter in the frame separately. In contrast, subjects could check all items simultaneously in the different category situation to see if any were numbers. They were able to perform this processing simultaneously because the process was automatic.
π‘ Why this matters: This experiment provides direct evidence that automatic processing does not consume attentional capacity, while controlled processing is capacity-limited and slows down with increased load.
Implications Schneider & Shiffrin argued that detecting numbers among letters didnβt place any load on the capacity and that is why subjects were not affected by frame size. Detecting letters among letters, however, was a hard task which placed a lot on attention capacity, which was affected negatively as frame size increased.
β Key Takeaways
Automatic processing requires minimal attention, occurs without intention or conscious awareness, and does not interfere with other mental activitiesβbut can lead to errors and poor recall. Automaticity is best viewed as a matter of degree, not a distinct category, as demonstrated by Morayβs shadowing expertise. In Schneider & Shiffrinβs visual scanning experiments, detecting a number among letters (different category) was automatic, yielding 80ms RT with no frame-size effect, while detecting a letter among letters (same category) required controlled processing, yielding 400ms RT and performance that deteriorated with larger frame sizes. The critical finding is that automatic processes place no load on attentional capacity, whereas controlled processes demand conscious attention and are capacity-limited.
π§ Quick Revision Questions
- What are the three criteria for a skill to be considered automatic, according to Posner and Snyder (1975)?
- In Schneider & Shiffrinβs experiments, what were the reaction times and accuracy rates for the different category condition vs. the same category condition?
- Why did frame size affect performance in the same category condition but not in the different category condition?
- How does Shiffrin & Schneiderβs perspective on automaticity differ from viewing it as a distinct category?
- What is the key difference between automatic processing and controlled processing in terms of attentional capacity?
π Lecture 12 β Automaticity (continued)
π Overview: This lecture continues the exploration of automatic processing by examining Shiffrin and Schneider's (1977) experiment demonstrating how practice can make discrimination tasks automatic. It introduces Hasher & Zacks' (1979) five criteria for distinguishing between automatic and effortful processes, providing a framework for understanding how different cognitive processes function and are affected by various conditions.
ποΈ Topics Covered
The lecture begins with Shiffrin and Schneider's second experiment showing automatic processing after 2100 trials with specific letter sets. It then examines the implications of automaticity for attention and multitasking. The main body introduces Hasher & Zacks' five criteria for automaticity: intentional vs. incidental learning, effects of instruction and practice, task interference, depression or high arousal, and developmental trends. The lecture concludes with a summary table comparing automatic and effortful processes across all five criteria.
π Lecture Summary
Experiment
Shiffrin and Schneider (1977) conducted another experiment where the target always came from one set of letters (B C D F G H J K L) and the distracters were from another set (Q R S T V W X Y Z). After 2100 trials, subjects achieved the same level of accuracy and reaction time as the different condition in the previous experiment. This demonstrated that subjects needed 2100 trials of practice before discriminating between two different sets of letters had become as automatic as discriminating numbers from letters.
π Example: In the experiment, after 2100 trials, subjects could identify if a letter was a target (from set B-L) or a distractor (from set Q-Z) automatically.
- Results: Reaction Time = 80 ms, Accuracy = 95%
π‘ Why this matters: This shows that automaticity is not just about whether two categories are different (like numbers vs letters), but also requires extensive practice to achieve with similar categories.
Implications
The results demonstrate that processes can become automatic with enough practice. When they do, devoting attention to them is no longer necessary. Performance is no longer affected by the number of processes being performed simultaneously.
Five Criteria for Automaticity
Hasher & Zacks (1979) proposed five criteria to distinguish between automatic and controlled or effortful processes. They also made predictions based on these five criteria.
1. Intentional vs. incidental learning
Intentional learning occurs when we are deliberately trying to learn; incidental learning occurs when we are not deliberately trying to learn. For example, teachers tell children to do things but do not do them themselves, or parents tell children not to lie but lie themselves β children learn from parents' and teachers' actions rather than their words.
π Definition β Incidental learning: Learning that occurs without deliberate intention or awareness of learning.
Incidental learning is as effective as intentional learning for automatic processes but is less effective for effortful learning. For example, we know that in Urdu the letter "seen" occurs more often than the letter "zhe" without trying to learn this information.
2. Effect of instruction & Practice
Instructions on how to perform a task and practice on the task should not affect automatic processes because they can already be carried out very efficiently. For instance, an expert cricketer who comes to the ground and the coach says "when you see the ball, hit it" β this instruction is not effective because the expert already knows what to do.
Both instruction and practice should affect effortful processes. Practice helps in learning well.
3. Task interference
Automatic processes should not interfere with each other because they require little or no capacity.
Effortful processes require considerable capacity and should interfere with each other when they exceed the amount of available capacity.
4. Depression or High arousal
Emotional states such as depression or high emotional arousal can reduce the effectiveness of effortful processes. If we are in a sad mood and someone gives us a difficult and demanding task, we cannot concentrate on that task and cannot learn well. We are not able to learn and pay attention in the classroom when we are in a sad mood.
Automatic processes should not be affected by emotional states. For example, if we have to brush our teeth, we can do it even if we are in a sad mood.
5. Developmental trends
Automatic processes show little change with age. Once a task is practiced, then age does not matter. They (most of the automatic processes) are acquired early and do not decline in old age.
Effortful processes show developmental changes. They are not performed as well by young children or the elderly. There are many things that old people cannot do because these tasks are not practiced. For example, if we are teaching math to an old man, he cannot learn well and easily.
We have to pay attention to all tasks β attention and practice make things automatic. For example, a student who has the habit of reading can succeed even if he is not intelligent.
Summary Table: Differences between Automatic and Effortless Processes
| Criterion | Automatic | Effortful |
|---|---|---|
| Intentional vs. incidental | No difference | Intentional better |
| Effect of instructions and practice | No effects | Improve performance |
| Task interference | No interference | Interference |
| Depression or high arousal | No effects | Poor performance |
| Developmental trends | None | Poor performance |
β Key Takeaways
The lecture establishes that automaticity develops through extensive practice, as demonstrated by Shiffrin and Schneider's experiment requiring 2100 trials for letter discrimination to become automatic. Once processes become automatic, they require no attention, show no interference with other tasks, and are unaffected by emotional states or age. The five criteria from Hasher & Zacks provide a systematic framework for distinguishing automatic from effortful processes: intentional vs incidental learning, effects of instruction and practice, task interference, emotional arousal effects, and developmental trends. The fundamental difference is that automatic processes operate efficiently without capacity limitations, while effortful processes require capacity, show interference, and are affected by instructions, emotional states, and age. For exam purposes, remember that automatic processes show no difference between intentional and incidental learning, no effects from instruction/practice, no task interference, no effects from depression/arousal, and no developmental changes.
π§ Quick Revision Questions
- How many trials of practice were needed in Shiffrin and Schneider's (1977) experiment for letter discrimination to become automatic, and what were the final reaction time and accuracy results?
- According to Hasher & Zacks' first criterion, how does incidental learning compare to intentional learning for automatic versus effortful processes?
- Why do automatic processes not interfere with each other, while effortful processes do interfere?
- How do emotional states like depression or high arousal differently affect automatic and effortful processes?
- What are the five criteria proposed by Hasher & Zacks for distinguishing between automatic and effortful processes?