An EEG Study Of The Neural Mechanisms Of Attention Using Natural Stimuli

The natural environment is dynamic and complex.The brain relies on the attention mechanism to allocate neural resources to preferentially process more important information.Therefore,investigating the neural mechanisms of attention is important to understanding the cognitive function of the brain.Most previous studies investigate attention mechanisms using artificial auditory or visual stimuli.Such highly controlled stimuli,however,are very different from stimuli in real life,and are therefore of low ecological validity.How attention modulates neural encoding of natural stimuli is an important question,but lacking investigation methods.Here,we proposed stimulus tagging techniques to quantify the neural responses to natural stimuli(including speech and videos)using electroencephalogram(EEG).With the stimulus tagging techniques,we investigated how bottom-up stimulus-driven attention and top-down goal-driven attention affect the neural responses to natural auditory/visual stimuli.The main contributions of this thesis are listed as follows.In the first study,we designed structured speech materials to quantify the neural responses to syllables and words when listening to semantically coherent spoken narratives,and investigated how top-down attention modulated the neural responses to syllables and words.The structured speech was either synthesize to remove prosodic boundaries of words or naturally read,which included natural prosodic boundaries of words.Results showed that when listening to the structured speech,cortical activity tracked both syllables and words,and the neural response to words was more strongly modulated by top-down attention than the neural response to syllables.In the second study,we introduced word-rate acoustic rhythms to the structured speech,and investigated how the bottom-up acoustic cues modulated the neural response to words.Results showed that the neural responses to acoustic cues and words were both observed,but with different spatial distribution.When participants attended to speech,cortical activity primarily tracked words.When participants' attention was diverted away from the speech,however,cortical activity primarily tracked the wordrate acoustic rhythm.Furthermore,the word-rate acoustic rhythm also enhanced the neural response to words,indicating that the bottom-up acoustic cues facilitated neural processing of words.In the third study,we proposed a visual stimulus tagging technique to quantify the neural responses to natural videos,and investigated how top-down and bottom-up attention modulated behavioral performance and neural responses during visual detection tasks.The stimulus tagging technique dynamically modulated the visual features(e.g.,luminance,contrast,etc.)of different areas in videos.When participants watched the modulated videos,the neural responses encoding different areas could be measured.In visual detection tasks,the behavioral performance and neural response to visual targets were affected by both top-down and bottom-up attention.In contrast,the neural response to the peripheral background was largely insensitive to top-down attention but affected by bottom-up attention.In the presence of a video background,both the responses to the targets and the peripheral background could predict individual behavioral performance.These results suggested that neural processing of visual targets and background jointly contributed to individual detection performance.In sum,we proposed a series of stimulus tagging techniques to characterize cortical processing of natural speech and videos.The study found that bottom-up and top-down attention interacted and jointly contributed to cognitive processing of audio and visual information in natural scenes.