LIMITED
CAPACITY AND ITS CONSEQUENCES FOR PROCESSING INFORMATION
A.) Selectivity: we choose
from many inputs which to attend to--if there are too many things to attend
to something suffers.
B.) Mental Effort or "Concentration":
energy or effort expended in processing a task.
Influenced by:
1. General arousal
2. Automatic vs Deliberate processing
3. Single vs Multiple Resource Theory: two tasks may tap different resource
pools
THEORIES
OF ATTENTION
There are two major categories of theories:
filter theories and capacity theories.
A.) Placement of an attention
filter--need to select out which incoming stimuli are attended to and which
are ignored.
1.) Early Selection Theory: Selective filter prior to abstract STM (prior
to Central Processing Unit, CPU) e.g., Broadbent, 1957; Triesman, 1960.
Target stimuli are processed more fully because there is perceptual suppression
of nontarget stimuli.
Multiple inputs are registered and held briefly
but only one message is analyzed perceptually--perceptual analysis requires
attention, which is limited to one input at a time.
ATTENTION IS A BOTTLENECK (the point at which
information overload is reached) THAT LIMITS PERCEPTION.
a.) Broadbent (1957): subjects attend to only one
ear--they perform well, but remember little about what they heard in the
other ear.
Selective filter occurs prior to abstract STM. Subjects can tell sensory
characteristics from the unattended ear: male/female voice; music/ noise/speech--no
semantic processing
A second example is the split span experiment:
(why not use 7?)
Right ear hears: 149
Left ear hears: 325
Numbers are presented simultaneously to both
ears, rapidly. Subjects can attend well to one ear at a time and must hold
the information from the other ear in a buffer memory.
If subjects are asked to recall the digits
ear by ear they will do well, but if asked to recall the digits by pairs
they do poorly.
This is the ‘funnel’ model in which movement
of a flap (1) uses energy; and (2) takes time--so with pair by pair retrieval
the buffer decays because the flap moves more times than with ear by ear
retrieval.
Broadbent's theory of strict serial processing
explains apparent parallel processing (watching TV and eating popcorn at
the same time) as time-sharing or multiplexing--two or more
tasks are done by simply switching attention back and forth between them.
Norman, 1969: Interrupted subjects periodically
during a shadowing task.
Found subjects were able to recall the last
1-2 words from the unattended ear.
Therefore all the information was processed
to some extent--
Filter had to occur after abstract STM.
Gray & Wedderburn (1960) found that if
you give the following:
Left Ear: then he
said what 4 hell
Right Ear: 3 2 5 1 the 9
subjects jump back and forth between ears.
Thus, it seems that messages from both ears get into sensory memory, and
subjects choose certain features, including semantic, for selecting what
to attend to in sensory memory.
This suggests the information in the unattended
ear is processed somewhere and leads to:
2.) Late Selection Theory: selective filter after
abstract STM; e.g., Deutsch & Deutsch, 1963; Norman, 1969.
The differential processing accorded to target
and nontarget stimuli is thought to be nonperceptual in nature.
According to this model all perceptual information
enters the system, but only that which is selected by the special attention
mechanism reaches higher processing centers.
The basic difference between early and late
selection theories lies in their view of the processing stage at which
unimportant aspects of information or stimuli are screened out.
Early-selection theories propose that certain
stimuli are never processed due to perceptual suppression of nontarget
stimuli.
Late-selection theories propose that the unimportant
information enters the system but simply is not chosen for further processing.
Triesman, 1969: suggested the filter can occur
at several places.
It is selective across a whole continuum of
input characteristics, and its tuning at any moment is determined by the
character of the information that has most recently entered consciousness.
The filter never completely blocks any incoming
stimulation; it simply attenuates (lessens in intensity) it, making is
more or less perceptible.
Incoming information may set the filter.
Sometimes called a ‘leaky’ filter model.
B.)
Capacity Allocation Theories - There is thought to be a pool of resources
which can be brought to bear and a central allocation mechanism which dumps
resources in one direction.
Example: walking and talking: if you become
interested in the conversation walking slows.
1.) Single-Resource Theory - one large single pool of processing capacity
(e.g., Kahneman, 1973).
2.) Multiple-Resource Theory - different pools for different tasks (e.g.,
Navon & Gopher, 1979)--i.e., hemispheres and dual tasks.
C.)
Kahneman’s model: suggested real world tasks require concurrent operation
of perceptual and cognitive processes.
Core Assumption: mental processes compete
for a single, limited pool of resources, or capacity.
He assumes there is also an allocation
policy which is affected by enduring dispositions (individual
differences in selection), momentary intentions (temporally-tied
selection), and evaluation of demands on capacity (subjective decision)
and this is all modulated by arousal.
D.)
Multimode Theory: Johnston & Heinz (1978)
Suggested both filter and capacity models
may contribute to actual attention processes.
We adapt to mode of attention best suited
to current task demands.
Paradigm: dual task methodology with 2 groups
Group One: shadow one ear based on pitch of
voice (male/female)--early filter task.
Group Two: shadow words based on category
membership--late filter task.
Concurrently press a button whenever a light
comes on.
Predictions:
Shadowing based on pitch takes less effort
and dual task will show relatively fast RTs.
Shadowing based on semantic process takes
more effort and dual task will show relatively slower RTs.
Outcome:
Supported predictions:
Conclusions:
Attention requires capacity.
Capacity requirements increase from early
to late filter placement.
AUTOMATIC
VS CONTROLLED (DELIBERATE) PROCESSING
A.)
Schneider & Shiffrin, 1977 investigated the ways which subjects scan
a visual display.
Two conditions: same-category, wherein
letters were scanned for on a field of letters, and different-category,
wherein a number was scanned for on a field of letters.
The same category condition was more difficult
for subjects--they required more time and made more errors.
S & S argued that before coming into the
lab subjects were well practiced at detecting a number among letters so
that this process was automatic.
In contrast, when subjects had to identify
a letter among letters deliberate processing was needed. Subjects had to
attend to each letter in each frame.
In the different category condition all items
could be checked simultaneously because the detection process was automatic.
In a second experiment they used the same
pool of letters as targets (e.g., b, c, d, f, g, h, j, k, l) and the same
pool of letters as distractors (q, r, s, t, v, w, x, y, z).
With practice subjects were able to learn
to learn this discrimination to the point where their performance became
automatic. Attention no longer needed to be devoted to scanning each individual
letter sequentially.
Similarly, LaBerge & Samuels (1974) taught
subjects to discriminate real letters and artificial letters. Across time,
subjects were able to discriminate mirror-image from identical letters.
B.)
Posner & Snyder (1975) concluded that automatic processes:
Occur without intention.
Require little attention.
Are well-practiced.
Complete themselves without conscious control.
Lower level processes are more likely to become
automatic than are later, more cognitive processes.
Do not interfere with one another.
C.)
Hasher & Zacks (1979, 1984) provided evidence we automatically encode
certain events.
In a series of studies across several age
groups they showed that information for frequency of occurrence of an event,
for its spatial features, and for temporal information is all automatically
encoded along with the target information.
Predicted effects for automatic
and effortful processing
Automatic ProcessingEffortful Processing
Intentional versus incidental learning:
No difference
Intentional Better
Effect of instructions and practice:
No effects
Both improve performance
Task interference:
No interference
Interference
Depression or high arousal:
No effects
Decreased performance
Developmental trends:
None
Decreased performance
in young children or elderly
PROCESS
SPECIFIC APPROACH: PRACTICAL APPLICATION
Current models of attention fail to adequately
address the clinical phenomena of attention deficits, or their remediation.
The process specific approach views attention
as the capacity to focus on particular stimuli over time and to flexibly
manipulate the information.
Attention is conceptualized as a multi-dimensional
cognitive capacity fundamental to information processing.
Deficits in memory and learning are often
a consequence of impaired attentional processing.
Within this model there are four levels or
components of attention. These are:
A.
Sustained Attention - Vigilance
The ability to maintain a consistent behavioral
response during continuous or repetitive activity.
B.
Selective Attention - Focused
The ability to maintain a cognitive set which
requires activation and inhibition of responses dependent upon discrimination
of stimuli--includes the ability to screen out extraneous visual or auditory
information.
C.
Alternating Attention - Switching
The capacity for mental flexibility which
allows for moving between tasks having different cognitive requirements.
D.
Divided Attention
The ability to simultaneously respond to multiple
tasks.