- Title
- An order driven bias in mismatch negativity amplitude data revealed through the multi-timescale sequence
- Creator
- Mullens, Daniel
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2017
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- The auditory system is capable of extrapolating repetitious patterns from the acoustic environment to form prediction models, which specify the most likely characteristics of subsequent sounds. When an encountered sound deviates from expectation, an auditory evoked potential (the mismatch negativity, or MMN) is elicited. Building on the studies of Todd et al.,(2011; 2013), this thesis employed the multi-timescale sequence to reveal a bias that distorts the impact of local sound statistics on MMN amplitude. By presenting alternating patterns of sound where a high probability repetition is interrupted by a rare physical deviant either rapidly (every 0.8 minutes) or slowly (every 2.4 minutes), MMN amplitude was shown to be modulated by more than the size of the deviation and confidence or stability in the model. This thesis through a series of studies shows instead that MMN amplitude is modulated by a strong first-impression bias that outweighs transition statistics, as a larger MMN is evident to sounds first encountered as a deviant in the slow compared to rapidly changing sequences, but no difference is evident for the sound first encountered as a standard. The thesis explored this bias over two parts, Part 1 explores stimulus salience (defined by experimentally-equated behavioural salience and natural salience) and Part 2 explored the mechanisms by which the bias manifests. Part 1 is comprised of two studies. Study 1 presented two short/long tones at equal probability within a go-stimulus detection task prior to the multi-timescale sequence. Participants were divided into groups and instructed to respond to the shorter tone with a button push or the longer tone with a button push. The MMN results within the multi-timescale sequence showed the bias was initially abolished but eventually developed for the go-stimulus assigned. This demonstrates the long-term impact of prior learning on deviance detection, and that even when prior importance/equivalence is learned, the bias ultimately returns. Study 2 presented participants with two phonemic /a/ and /i/ sounds within the multi-timescale sequence. The results indicated that the first-impression responsiveness bias reflected in MMN amplitude to pure tones was replicated in phonemic sound presentations. Taken together, the results of Part 1 suggest that a sounds exposure, order or role (at first-impression) biases future responsiveness, outweighing local sound statistics and modulating MMN amplitude well beyond the limits of sensory memory. Part 2 is comprised of three chapters and explored the mechanisms of the bias and how it relates to second-order learning. Study 3 explored how MMN amplitude changes as a function of role stability within sequences of the multi-timescale structure by re-analysing the data of Todd et al., (2013). Comparison between MMN amplitude size at transition points in the sequence (first-half; immediately after change in role) versus later in a block (second-half; when roles are established) revealed a processing bias at transition points only, with this difference dissipating over the course of the entire block (no interaction or effect in second-half data). Further, the results indicated that the difference in tone processing is reactivated each time a new block starts. The results expose a bias consistent with high confidence in predictions when subsequent sequence blocks match initial statistics but low confidence for blocks containing those contrary to that first encountered. Study 4 and Study 5 showed similar first-half data trends in re-analyses of Study 1 and Study 2 respectively. Subsequently, Study 6 and Study 7 was designed to test if a processing bias could be attributed to assumptions about the tone roles (predictable versus unpredictable) in the first encountered context, weighted by the stability of that context. Study 6 showed that this bias is initially prevented if there is no 1:1 mapping between sound attributes and roles, but it returns once the auditory system determines which properties provide the highest predictive value. Study 7 showed that confidence in this bias drops if assumptions about the temporal stability of the pattern are violated. Both Study 6 and Study 7 provide evidence that the auditory system extrapolates patterns on multiple timescales to adjust response to prediction-errors, but profoundly distorts transition-statistics due to assumptions formed on the basis of first-impressions.
- Subject
- MMN; multi-timescale sequence; first impressions; primacy bias
- Identifier
- http://hdl.handle.net/1959.13/1337618
- Identifier
- uon:27879
- Rights
- Copyright 2017 Daniel Mullens
- Language
- eng
- Full Text
- Hits: 665
- Visitors: 943
- Downloads: 249
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 6 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 117 KB | Adobe Acrobat PDF | View Details Download |