Wk 3: visual search + attentional bottlenecks

3a

visual search

  • aspect of spatial attention

  • involves shifts of attention

  • visual search efficiency can also impact attentional breadth

terms:

  • serial processing

    • one process commences only when the previous is finished

    • sequential

  • parallel processing

    • multiple processes occur simultaneously

  • dimension

    • complete range of variation which is separately analysed by some funcitonally independent perceptual system

    • e.g. colour, orientation

  • feature

    • a particular value along a dimension

    • e.g. red, vertical

  • binding problem

    • putting together dimension and feature into something meaningful

feature integration theory

  • pre-attentive processing of single features

    • basic visual features are processed automatically (without focused attention) and in parallel across the visual field. (e.g. colour, orientation)

    • features are encoded in separate “feature maps”

    • analogy: photoshop, separate layers for colour, edges, motion blue. each layer highlights different aspects of the image. to understand the full object, you need to focus on a location and combing layers. attention=glue

visual search array

  • targets: searched-for

  • other items: distractors or non-targets

  • common dv: RT to detect or identify the target

  • set size (total no. of items) allows us to infer whether processing is serial or parallel

  • set size increase=long processing for serial RT

  • set size change and not processing time change = parallel processing

  • search for:

    • single feature (i.e. just red) = parallel

    • conjunction of features (i.e. red and square) = serial

      • RT increases with set size and target-absent trials

feature vs conjunction search

  • modern: but how similar are distrators from each other

  • modern: degrees of efficiency, rather than serial vs parallel

Guided search

  • search is guided by a priority map

  • what you search for previously, influences what you’re searching for currently

  • sources combine to form priority map:

    • top-down - goals and expectations (e.g. I’m looking for a red book)

    • bottom-up - salient features in the env (e.g. brigh colours, sudden motion)

    • prior history

    • reward - learned associations with value

    • scene syntax and semantics - knowledge of how scenes are typically structured

applied context of visual search

  • diagnostic medical imaging

  • baggage screening

humans are much more likely to miss the same target when it occurs in a low-prevalence context than when the same targeam occurs in a high-prevalence context.

  • i.e. we are more likely to miss a target when it’s chance of appearing is 1%

  • vs target has a chance of 99% of appearing

questions

  • how do you think that visual search efficiency impacts attentional breadth?

    • what attentional breadth would a parallel (more efficient) search induce?

    • what attentional breadth would a serial (less efficient) search induce?

3B: attentional bottlenecks

  • attention across time.

attentional blink (AB)

  • a brief period of impaired perception

  • occurs after detecting a target in a rapid stream of visual stimuli

  • a “blink” in attention

experimental setup:

  • rapid serial visual representation (RSVP)

  1. items flash one at a time at the same location every ~100ms

  2. participants are asked to detect 2 targets in the stream

  3. when you detect T1, resources is taken up

  4. T2 appears ~200-500ms after T1, you’re more likely to miss this

conditional accuracy

  • T1 can be reliably detected, accuracy high

  • T2 accuracy low

  • therefore, temporal bottleneck (limitation in cognitive processing capacity over time)

Stimulus Onset Asynchrony (SOA): time interval between the appearance of T1 and T2

  • short SOA = bad T2 accuracy (brain still processing/blink period)

  • long SOA = T2 accuracy recovers

Is the AB an attentional or sensory limitation in processing?

  • T1 presence influences T2 accuracy

  • hence it is is in attentional limitation, not perceptual

AB: two-stage model

  1. stimulus detection, holds multiple items but fragile and temporary representation

  2. encoding into a durable, explicity, conscious representation (e.g. short-term memory)

  • encoding is slow and creates a “bottleneck”

  • T2 accuracy bad due to encoding stage (stage 2)

how does emotion impact temporal attention

emotional attentional blink

T1 and T2 are both emotionally-salient

  • threatening stimuli as T2 are less affected by the blink, especially with high anxiety traits

  • threatening stimuli as T1 can increase the blink, only when participants are asked to judge the emotion (task relevance matters)

emotion-induced blindness (EIB)

T1 and distractor, both emotionally salient

  • also involves RSVP of images in a single spatial location

  • lag 2 ~200ms after distractor target appears, accuracy low

  • lag 8 ~800ms after distractor target appears, accuracy high, similar to emotionally-neutral