Content, granularity, and type 2 sensitivity of subjective measures of visual consciousness.
Dissertation, LMU München: Graduate School of Systemic Neurosciences (GSN)
According to several major theories in the field of consciousness research, the valid assessment of conscious awareness requires subjective measures, i.e. participants’ reports about their conscious experience. However, there is a considerable amount of uncertainty in the field if and how scientifically valuable data can be obtained from subjective measures.
The present work empirically examines how subjective measures of conscious awareness need to be designed and applied to provide maximally useful data for empirical studies of visual consciousness. Specifically, it is investigated what contents subjective measures should require participants to report, at which granularity subjective measures ought to be recorded, and what statistical procedures should be used to quantify the relation between subjective measures and discrimination task performance.
Concerning content, subjective measures that referred to the accuracy of a preceding discrimination response and subjective measures referring to participants’ visual experience of the task-relevant stimulus feature were compared during a series of visual psychophysical experiments. Subjective measures about the accuracy of the responses were associated with more liberal psychophysical thresholds: At lower stimulus quality, participants reported that they feel confident that their discrimination response was correct without reporting a visual experience of the stimulus feature. Only at greater stimulus quality, they reported that they had a visual experience of the stimulus feature in addition to being confident. Moreover, subjective measures about confidence in discrimination responses predicted task accuracy more efficiently than measures about visual experience. Finally, subjective measures of experience and task accuracy as content were compared while event-related potentials (ERP) were recorded. The earliest electrophysiological correlates of subjective measures where predictive of the fact if participants reported that they selected the response to the discrimination task based on knowledge instead of guessing, but were not yet predictive whether participants reported a clear experience over and above making the task response based on knowledge. The strongest ERP correlate of visual experience occurred a short period in time before participants responded to the discrimination task. As a consequence, it is argued that conceptual considerations are required which conscious contents are relevant for a specific research question, and subjective measures should be about the relevant contents accordingly.
Concerning the granularity of subjective measures, a continuous scale and a scale with four discrete labelled categories were compared as subjective measure of conscious experience of motion. The subjective measures contained more information when participants used the continuous scale instead of the discrete scales. The greater amount of information provided by continuous scales rendered subjective measures more predictive of task accuracy and enhanced internal consistency.
Regarding the statistical procedure to quantify the relation between subjective measures and task performance, it was found that logistic regression is a suboptimal method because the relationship between subjective measures and the transformed accuracy was frequently not linear. In contrast, meta-da, a measure of the relationship between subjective reports and task accuracy derived from signal detection theory (SDT), provided the most consistent results across all studies.
Overall, it is concluded that subjective measures are suited to provide highly useful data to address non-trivial research questions for the scientific study of consciousness: As prerequisite, the content of a subjective measures should be tailored to the current research question. In addition, the problem of a lacking objective standard can be addressed by using the relation between subjective measures and task performance as a reference frame.