TB6 P&C Tutorial; Change Blindness (Fenn et al)
- Created by: mint75
- Created on: 22-10-15 14:47
View mindmap
- Fenn et al (2011) Change deafness in a telephone conversation
- Experiment 1
- To test whether listeners would detect changes in talker during a normal phone conversation
- Found that only 1 pp detected the change.
- In the voice recognition test (15/16 pps), there was no preference for either experimenter voice (60/40%)
- Results suggest that even whilst actively attending to the message, listeners often fail to detect a change in talker voice.
- Potentially support the idea that a 'gist'/general memory of voices is encoded.
- To test whether listeners would detect changes in talker during a normal phone conversation
- Experiment 2
- To test memory for a voice encountered in the conversation compared to a novel voice
- Testing whether experiment 1 results were due to chance. (If pps had no general memory of either voice they should NOT be able to discriminate between the heard and novel voice).
- The results suggest that the pps DID encode/retain some general information about the heard voice, enough to significantly distinguish from the novel voice.
- These findings are also consistent with change blindness research which states that both voices will be encoded but change will NOT be automatically detected.
- The results suggest that the pps DID encode/retain some general information about the heard voice, enough to significantly distinguish from the novel voice.
- Found that only 1/24 pps noticed the voice change.
- In the voice test, 74% of pps who did not notice the change manage to correctly identify the the experimenter rather than the novel voice
- Testing whether experiment 1 results were due to chance. (If pps had no general memory of either voice they should NOT be able to discriminate between the heard and novel voice).
- To test memory for a voice encountered in the conversation compared to a novel voice
- Experiment 3
- To test whether pps could detect the change in voice after a warning that 50% of pps will have 2 talkers, even though 100% actually did.
- Testing the role of expectation on change deafness
- Found that 75% of participants detected a change and the majority of these interrupted the experiment at the correct place
- This is a SIGNIFICANTLY higher detection rate than both 1 and 2
- Because the conversation clearly did not impose a cognitive load that prevented change detection, these results suggest that listeners do not normally expect a talker change and so do not moniter at an analytical level.
- When this expectation is suspended (with the warning), change detection increases; expectation may influence how listeners interpret voice detection.
- To test whether pps could detect the change in voice after a warning that 50% of pps will have 2 talkers, even though 100% actually did.
- Experiment 4
- To test the effect of expectation on talker change detection with the actual change occurrence manipulated
- The same methodology as 3 except that 100% of pps were expecting a talker change even though it actually only happened to 50% of them.
- All but one participants in the group that changed talker detected the change.
- In the same talker group, many pps were confused and 'assumed' there was a change without being able to concretely confirm
- These results again confirm that the way listeners attend/encode information is influenced by expectation
- To test the effect of expectation on talker change detection with the actual change occurrence manipulated
- Experiment 5
- To investigate what sort of changes pps spontaneously notice
- Changed the gender of the talker
- Found that 11/12 pps reported the gender change
- Significantly more detection than any other group
- Concurrent with Cherry (1953) in that significant elements are still noticed even when unattended
- Significantly more detection than any other group
- General evaluation points
- Strong theoretical background to findings, concurrent with earlier research which does NOT suggest automatic encoding/ detection
- Small sample sizes (n=12) used, small external validity/ statistical power
- e.g only non-parametric tests were used, may see a significant result when there isn't
- A control needed for male/female ratio? Female pps may find it easier to identify female voices
- Should have used female/female, male/male, male/female?
- Well controlled; conditions were counterbalanced andthe same voices were often used to reduce extraneous acoustic variables
- Controlled for hearing/ speech impediments
- Although necessary as not to inform the pp of the study, time/day may also be important. If pp is working then may be paying less attention to researcher
- Experiment 1
Comments
No comments have yet been made