![]() ![]() The dataset includes 22,248 images related to 1,854 concepts. This approach was employed in a recent open source dataset that captured EEG measurements for object recognition using a rapid serial visual presentation paradigm 2. To drive decoding of a class based on semantic information, stimuli must vary in their low level sensory details. For example, the differences between imagery and vision appear to be most pronounced in the early visual cortex, with greater overlap occurring higher up in the visual hierarchy 21 and at time points linked to high level perceptual processing 14. Efforts are being made to determine the spatiotemporal extent of these shared neural representations 18, 19, 20, which may be most invariant in brain regions and time points associated with latent representations i.e. This task invariance has enabled cross-decoding between perception and imagination 16, 17. Growing evidence of neural overlap between perception and imagination 14, 15 may also facilitate generalisability. This can help increase robustness to real world data heterogeneity. Invariance to low level sensory detail can be considered a desirable quality in BCI systems in which within class generalisabilty is a key goal. The advantage of decoding semantic information, as opposed to sensory based information such as visual details, is that semantic representation is partially invariant across modalities 9, 10, 11, 12, 13. ![]() For example, was the object in an image shown to an observer a flower or a guitar? The low level visual and auditory sensory details of the semantic concept, such as whether the flower is yellow or purple, are ignored with a focus on the high level meaning of ‘flower’. In contrast, semantic decoding extracts conceptual information such as object types or classes. Visual decoding involves decoding simple low level visual components such as colour and shape, or complex naturalistic images of objects, scenes and faces. However, the lack of open source EEG datasets for decoding imagined and perceived semantic level information is hindering progress towards this research goal. Recently, there has been growing interest in decoding alternative information forms such as auditory and visual, perception and imagination 7, and semantic information 8. These datasets enable the development of sophisticated techniques for analysis and decoding, which can be used to investigate neural representation mechanisms and improve decoding performance for EEG based BCIs.ĭifferent paradigms have been used for EEG based BCIs such as Event Related Potential (ERP) BCIs for decoding inner speech 1, 3, Steady-State Visual Evoked Potentials (SSVEPs) 4 and motor imagery 5, and oscillatory activity driven BCIs for tasks such as drowsiness detection 6. For example, EEG datasets for inner speech commands 1 and for object recognition 2 were recently created and shared to address a lack of publicly available datasets in these areas. A single open source dataset can form the basis of many varied research projects, and thus can more rapidly advance scientific progress. ![]() Although EEG datasets can be time consuming and expensive to obtain, they are extremely valuable. Similar content being viewed by othersīrain computer interfacing and cognitive neuroscience are fields which rely on high quality brain activity based datasets. Surface electroencephalography (EEG) is a popular choice of neuroimaging technique for BCIs due to its accessibility in terms of cost and mobility, its high temporal resolution and non-invasiveness. The aim is for the dataset to be open for purposes such as BCI related decoding and for better understanding the neural mechanisms behind perception, imagination and across the sensory modalities when the semantic category is held constant. Here we present an open source multisensory imagination and perception dataset, with twelve participants, acquired with a 124 EEG channel system. In particular, there is a scarcity of existing open source EEG datasets for imagined visual content, and to our knowledge there are no open source EEG datasets for semantics captured through multiple sensory modalities for both perceived and imagined content. These stimuli representations can be either imagined or perceived by the BCI user. The same semantic meaning can be conveyed in different representations, such as visual (orthographic and pictorial) and auditory (spoken words). A range of input representations has been explored for BCIs. Electroencephalography (EEG) is a widely-used neuroimaging technique in Brain Computer Interfaces (BCIs) due to its non-invasive nature, accessibility and high temporal resolution. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |