Module 4 - Attention Pt. 1

How do we decide to pay attention?

Automatic

Little effort to do

Selective attention

Select relevant info and ignore irrelevant

• Process less info from other tasks

• Novelties were distracting

• Intrusions from things meant to be ignored

• Context, syntax, storyline helped keep reader on track

Failure of selective attention 

• Present to receive info, but no access to it

• Mind wanders and don’t rmbr info

Hal Pashler

Awareness is only small portion of stimuli impinging on sensory system 

Actively focus on stimuli 

• Selective attention

Introspection

Think about experiences as you perform the task

Syntax

rules for putting sentences together

Key components of theoretical approaches to attention?

1. When selection occurs/how much processing do we do before we attend to it?

2. Fate of info selected 

3. Fate of unselected info 

Donald Broadbent 

Filter theory 

Early Selection model

Early selection model

Broadbent

Attentional selection happens BEFORE person knows what info is

• Very early on  

Based only on physical characteristics (i.e. source, pitch, volume, colour, brightness) 

Selection occurs before semantics (meaning) is processed 

Binaural Presentation/Dichotic listening task

Hear 2 diff things in each ear

Must focus on message (lots of mental effort) 

Less resources are available to process background message 

Shadow

repeating everything you heard as quickly and accurately as possible

Cherry’s study

Asked to report unattended message

• Ppl cound’t say whether it was speech of M/F or noise

• Couldn’t recall content of message or language used 

Wood & Cowan 1995

Used dichotic task (2 diff stories told)

Processing of unattended message worsened their performance on main task (more errors) 

Shifted to unattended message was unintentional (w/o awareness) 

• Detection of backwards speech interfered w/ task at hand. More errors, worse performance 

Conway, Cowan & Bunting 2001

Detecting your name in unattended message = longer working memory span

Lower working memory capacity = less able to block unattended message 

Low working memory = LESS focused 

Broadbent’s findings w/ Selective Attention

Limits to how much person can attend to 

• Didn’t notice anything about message in other ear

• CAN notice volume, pitch of message, but not content 

  • Even if it was the same word repeated over and over

• Didn’t notice language change

What is the attentional selection/filter?

Broadbent

Early Selection model

Attention = you select material from right channel and process it for understanding 

Filter based on physical aspect of attended message. Process these basic properties in unattended message 

• location of source 

• pitch/loudness

Filter chooses info to process early (BEFORE meaning is identified)

All unattended messages are filtered out (no processing of info)

Unselected was not proceed beyond basic features 

• Dunno what info is for 

Amount of info we process is limited 

• 2 messages w/ small info could be processed slowly @ same time 

Explains why unattended message wasn’t processed

Goal of Attentional Selection

Protects from overload

Problem w/ Broadbent Filter Theory

Cocktail Party Effect: @ loud party, you can still hear your name

Moray 1959 Experiment

Cocktail Party Effect

Dichotic listening task, notice own names in unattended channel 

Only important material penetrates filters 

• Analyze for meaning 

• But didn’t always hear name 

• Shadowing doesn’t take 100% of attention 

Attention will lapse and shift to unattended message (recognize name) 

Attenuation Theory

A.k.a Leaky Filter Model → Anne Treisman 1960 

Pick up info from unattended channel when its important to us 

• Word of importance: fire, names 

• low threshold recognized easily @ low volumes 

Info can leak thru filter and be processed if info has value to us

Volume turned down 

• Some meaningful info in unattended messages might still be available (not completely blocked out)

• Is attenuated/weakened

Messages w/ low thresholds can be recovered from unattended messages

Special/relevant info leaks into awareness and bypasses filter 

3 Types of Analyses of Attenuation Theory 

1. Physical properties: pitch, loudness 

2. Linguistic: parsing message into syllabus + words 

3. Semantic: meaning of message 

Priming of Words - Attenuation Theory 

Context of word can lower threshold 

• Little effort to hear and process it 

• Cat is primed “dog chased the…”

Hearing previous word primes to detect words that followed even when in unattended message 

• Only process enough to separate attended from unattended 

  • I.e. differ physical characteristics, only process to this level and reject unattended msg 

• Processing meaning = effortful (only done when necessary) 

Attenuation vs Filter Theory

Attenuation: Many diff kinds of analyses of messages. Unattended messages weakened, but info is still available 

Filter: only 1 message. Unattended messages once processed are discarded + blocked. 

Corteen & Wood 1972 

Selective attention

• Paired city names w/ electric shock

Unattended info was processed → Galvanic Skin Response

Process info to semantic level processing 

• Unattended info is processed to level of meaning, even if person isn’t aware 

Late Selection Theory

A.k.a Deutsch Norman Model of Attention

Deutsch & Deutsch

Selective attention goes to late stage to direct awareness and guide response 

All info (un/attended) processed to point of meaning in LTM 

• Attentional selection happens AFTER this processing 

• Unlikely unattended msgs processing for meaning 

Despite unaware of info, we process and activates its representation in memory 

• Influences behaviour 

Pashler’s Interpretation of Late Selection Theory

Minimum, recognize familiar objects

Unselective

No capacity limitations 

Don’t voluntarily choose what we recognize 

Similarities of Late Selection Theory & Filter Theory

Uses bottleneck analogy, but just later processing

All material is processed, but you judge importance 

• Importance determines elaboration (more likely to be retained) 

Importance Criteria

Context + personal significance (i.e. name)

Alertness 

• Sleeping: only very important msgs 

  • Baby crying 

Attentional system determines importance of incoming messages 

Model of Attention

Daniel Kahneman

Complex stimulus, harder processing, more attentional resources required

Humans control direction of mental resources

• what to focus on

Monthly budget metaphor 

• Factors influence allocation: 

  • Extent, type of mental resources available 

State of arousal & resource capacity?

More resources available to devote to tasks 

Arousal depends on….

Complexity of task

Easy = less aroused, less resources used 

Model of Attention

What does arousal affect? 

Capacity (sum of resources) 

Model of Attention

Allocation Policy Criteria

Affected by dispositions….

• Biases

• Momentary intentions 

• Evaluation of demands on their capacity 

• Our interests, what we find important, in the mood for 

Model of Attention

Attention = mental effort

More effort, more attention

Model of Attention

Data is limited

Depends on quality of stimulus

NOT mental effort/concentration

Model of Attention

Greater effort/concentration + performance 

Better performance on some tasks 

Resource limited processing

• Performance constrained by mental resource/capacity 

Pros Automatic Behaviour

Less attention needed to perform behaviour

It becomes more automatic

Requires less attentional resources to perform

Spend less attentional resources 

Capacity important determining variable that governs the # of things we can do simultaneously 

• Difficulty impacts capacity 

• Familiarity with task 

Con of Automatic Behaviour

Become so automatic that we can’t prevent it from happening

When we don’t want to do it 

Impairs what we want to do 

Stroop Effect

Interfering effects of relative automized process 

Automatized process interferes w/ behaviour  

Stroop Experiment

John Ridley Stroop

Difficult- reading is automatic. interferes w/ ability to name ink colour. Can’t prevent it, even when effects are deleterious.

• Automatic: no attention, can’t be inhibited 

Manipulate manner participant attends to Stroop stimulus 

• Reduces or eliminates Stroop effect 

• Automatic and not require attention, word reading depends on attention 

Colour doesn’t match the name - tendency to read the name

Stroop Interference

Reading is automatic

Interferes w/ ability to name ink colour

Can’t prevent it, even when effects are deleterious

Beginner Readers affected by Stroop Interference? 

Beginner readers

• Reading is controlled process 

• Young kids and new speakers not troubled by Stroop 

Schneider & Shiffrin 1977

• Controlled Processing 

Serial: 1 set of info processed @ time 

Requires attention 

Capacity limited 

• Multitasking, insufficient resources to do it 

Conscious control 

• Very aware @ the beginning 

Difficult task + unfamiliar process 

Varied mapping 

Automatic Processing Criteria

Posner & Snyder

1. Must occur w/o intention 

2. W/o conscious awareness 

3. Not interfere w/ other mental

i.e. Make turn without intending to during drive

Schneider & Shiffrin 1977 Experiment

Automation testing in lab

Frames: search for targets (letters/#s) in array of letters/#s 

-ve: target not in frame 

Easy to find when target is diff from array 

Distractors: non-target characters 

#distractors makes little to no difference if distractors are diff type from target 

Target + distractor same = #distractors DOES make a difference 

2 Conditions:

1. Varied mapping: mix # and letters

• Distractors become targets btwn trials

• Performance depended on all 3 variables (memory set, frame size, frame time)

1. Consistent mapping

• Only #s

• Distractors: letters

• Expected task: easier, less capacity

• Automatic process: performance only varied w/ frame time

• Accuracy depends: length time frame displayed

Varied: frame size, frame time (display duration), memory set (# of targets)

Automatic Processing

No attention required, automatic w/ practise D

No conscious awareness

No interference w/ other mental activities

• Results might interfere with other activities

Works in parallel (works while other processes are happening simultaneously)

Does not constrain capacity

For easy tasks

• Little effort or concentration required

• Several searches done simultaneously

Is human attention flexible?

Yes, we can manipulate where we allocate our attention

Divided Attention

Perform 2 tasks at once

Dual Task Performance

Spelke, Hirst, Neisser

• Students write words dictated while reading short stories 

• Test reading comprehension 

• Reading speeds similar btwn reading & writing and just reading 

• Process meaning w/o conscious attention 

Hypothesis of Spelke, Hirst, Neisser

Hypothesis #1: Alternate attention btwn 2 tasks

• Reading speeds comparable btwn 2 

• Alternate attention w/o lag? → REFUTED 

  • Recieved SAME TRAINING 

#2: 1 task performed automatically 

• REFUTED: required intentional and conscious awareness 

• Paid attention to task to process dictation for meaning 

#3: Combine 2 separate tasks  

• Practise increased efficiency 

Psychological Refractory Period

2 different tasks 

• No interference - can perform both tasks successively 

Work on 1st task, no attention to work on second 

Bank Teller Analogy

• Wait time for C2 depends on interval btwn arrival time and C1 

  • Can’t work w/ C2 until done w/ C1 

  • Wait time analogous to slowed response time to S2

  • Teller is bottle neck, limiting factor of speed for S2 processing 

  • Retrieve info from memory caused bottle neck and disrupts attention of 2nd task 

Mike Tombu & Pierre Jolicoeur 

Attention is all or nothing, but can be flexible/shared

Ppl can allocate some attention to demand of task

Capacity Sharing model

Share attention btwn 2 takes

Impaired performance on 1 relative to full attention

Attention Hypothesis of Automation

Attention needed during practise

Attention determines what gets learned/remembered 

Logan et al. 1996

Learning is result of attending

Attention affects what info gets encoded into memory and what can be retrieved 

Words consistency paired

Advantage in performance when SPECIFICS of target detection task forces attention to both words 

• No advantage from consistent pairings 

  • Recalled fewer distractor words 

Colour cues → easier to ignore 2nd word (no attention, no learning) 

Consistent practise, unlikely to learn words if no reason to pay attention 

Spatial Attention

Visual search: look for mom @ airport

Spatial cue: flash from camera

Spatial Cueing

Posner and Colleagues

Valid trial: Target in same location as cue 

• Fastest reaction 

• Performance facilitated on valid trails bc shifted attention to location of target BEFORE it appeared 

Invalid trial: Target in diff location as cue 

• Slowest, most errors 

• Cost to shift location to wrong location 

Neutral trial: 2 headed cues target equally likely to appear on either side 

Input Attention

Input Attention

Attention is spotlight that enhances efficiency of detection of events within its beam 

Makes easier for stimulus to be brought into system for processing 

Spotlight enhances efficiency of detection during search 

Visual Search

Treisman et al.

• Simple objects w/ differing features 

• Target pops out, background items aren’t

• Proof: detection of individual features are automatic

• Doesn’t require attention

• Performed in parallel

• All items can be searched @ once 

We recognizer individual features automatically (no effort)

Combination of Features/Conjunction Search

Search difficult

Response times varied based on # background numbers 

More distractors, longer time 

Higher controlled nonautomatic processing 

• Moving attentional spotlight 

Feature Integration Theory

Treisman & Schmidt

Perceive objects in 2 stages 

1. Pre-attentive/automatic: register feature of objects (shape, colour). Identify objects w/ the SAME feature w/o attention 

2. Glue features tgthr into unified object. Identify complex objects, or detect objects that share features w/ other background objects 

Rewuires mental effort (to recognize complicated objects) 

Illusory Conjunctions

Gluing errors due to attention overload or diversion of attention

• i.e. Combine 2 stimulus/erroneous/illusory 

Inattentional Blindness

Connected to change blindness

Not perceive stimulus right in front of you UNLESS you pay attention to it 

• Only perceive what you attend to (especially if unexpected event is dissimilar to focus of our attention) 

• Detecting target requires attention 

Ray Klein Findings 

Spatial cueing

Visual search task

Easier to detect target when appear in availability cued location than uncued location 

Inhibition of Return

Inhibition of Return 

Target appearance is delayed

Observers are slower to detect target @ validly cued location

• Adaptive fxn to enhance search 

• Inhibitory tagging system 

Keep track of locations searched to prevent search in same spot 

• Cue brings attention to location

• Longer delay + appearance of target, move attention elsewhere 

Target appears in cued location, observer is SLOWER to move attention back bc marked as search already 

General areas of brain responsible for attention 

RIGHT SIDE

• Frontal: selects motor response and develop plans 

• Posterior parietal lobe 

• Pre-frontal 

32 areas of brain active when process visual stimulus 

3 Areas of Orienting Network 

1. Posterior parietal lobe: Disengage from previous cue 

2. Superior colliculus: Move to new target 

3. Pulvinar: Enhance target @ location 

Focuses attention

Select info from sensory input + executive control network 

Posher & Raichle 

Operates independently 

ADHD

Common in M

Can’t sustain vigilance on boring, dull, repetitive tasks

Can’t inhibit ongoing response

Events Related Potential

Average EEG trails to lower noise (1ms after stimulus)

Amplitude larger for attended vs unattended stimulus 

Diff @ 80 ms after stimulus - time for info to go from sensory receptors in ear to cerebral hemisphere 

Effect in brain, not ear 

Response enhanced when presented in attended location