Serial Biases and Central Tendency in Time Reproduction

The research presented in this thesis investigates how temporal context influences the per-ception of time, specifically in a time reproduction task. Temporal context in our framework refers to the temporal proximity (recent history) to the current trail and represents the du-rations perceived and reproduced throughout the experiment, either on recent trials (local context) or on all the trials our participants were exposed to throughout the whole experi-ment (global context). Our studies aim to fill a gap in the existing literature by investigating the interactions between local and global contexts in a time reproduction task. The studies explore the connection between local and global biases, they try to understand the under-lying mechanism(s) and the brain correlates (i.e., the “where” in the brain and the “when” in information processing these biases take place). In Chapter 2, we explore the relation-ship between serial biases and central tendency. By manipulating the local context only, we show that serial biases contribute to central tendency and that both phenomena are best explained by a model with a dynamically updating prior. To form a percept, the model combines the observed duration with the prior expectation which is a weighted sum of pre-vious reproductions, with the most recent ones contributing more than others. In Chapter 3, we focus on the brain signature of serial biases as measured with Electroencephalography (EEG). In this chapter we try to dissociate the contribution of previous motor reproduction from that of preceding stimulus and we show that even though serial biases in reproduction task are largely driven by preceding reproduced durations, the effect is strongest when the reproduced times reflect the perceived durations. Additionally, we find a signature of the response-driven serial bias at a post-perceptual processing stage, in the spectral power in a wide alpha frequency range. Finally, in Chapter 4, we use an auditory reproduction task in a Magnetic Resonance Imaging (fMRI) experiment to show that the relationship between serial dependence and central tendency may be affected by the transition from one temporal context to another. Additionally, the fMRI data show that at the offset of currently perceived stimulus, changes of the hemodynamic activity in the thalamus and in the para-cingulate cortex are associated with changes in measures of global and local biases. Both these pieces of evidence point towards a post-perceptual stage of processing for serial biases. The evi-dence collected in our experiments show that local and global contextual biases interact with each other and that for serial bias in time reproduction the type of preceding action plays a key role. The thesis helps in bridging the gap between our knowledge about global and local biases in time perception and provides interesting results on the mechanisms, stages of processing and brain correlates.