脳の「内的ノイズ」研究が示すヒント——視覚トレーニングという新しい設計思想
脳の「内的ノイズ」研究が示すヒント——視覚トレーニングという新しい設計思想 「盲の人のほうが音程を正確に聞き分けられるのはなぜか」——Journal of Neuroscienceに掲載された新しい研究が、興味深い答えを示しています。彼らの聴覚系は、音をより一貫して安定して処理できる。つまり「内的ノイズが少ない」というのです。そして驚くべきことに、晴眼の音楽家も似たメカニズムで聴覚を高めている可能性があるといいます。 ここで注目したいのは「能力の高さ」そのものではなく、「訓練によって脳の処理の質が変わりうる」という示唆です。 知覚を「入力設計」として捉え直す…
A new study published in the Journal of Neuroscience suggests that blind individuals are better at discerning musical intervals due to their auditory system processing sound more consistently and with less internal noise. Surprisingly, even those with normal vision may use a similar mechanism to enhance their auditory perception. The key takeaway is not the "level of ability" itself, but the idea that training can significantly improve the quality of brain processing.
When viewed through an engineering lens, the study reveals an interesting structure. The accuracy of perception depends not only on the physical capabilities of the sensors (retina and ganglion cells) but also on the noise floor in the downstream neural circuits, specifically the signal-to-noise ratio (SNR). Training, therefore, involves improving this SNR through software, not just hardware.
This concept applies to visual perception as well. In cases like lazy eye (amblyopia) or astigmatism, the integration of information from both eyes plays a crucial role. More important than individual visual acuity is how well the brain fuses the two systems to create a single 3D image. The challenge lies in merging the two distinct input systems from each eye into one cohesive perception.
Amblyotube, developed by Seven Sports, is a visual coordination and attention training software for Meta Quest. It presents different images to each eye, creating a "visual gym" similar to YouTube content. The dominant eye sees a digital blur (Dominant Eye Shader), while the amblyopic eye receives AI sharpening (MFBF) that detects faces, outlines, and rhythmic flickers.
The brain integrates the magenta focus cue and the dominant eye's grayscale dots to form a single 3D image. The wedge occlusion separates each eye's viewing area, reducing competition and allowing for a seamless experience.
What's fascinating is that this approach doesn't require hardware like patches or red/blue glasses. It controls the software presentation of left and right inputs on existing VR headsets, using shader techniques and AI edge enhancement to create a difference in input. This design approach, focusing on software input rather than hardware constraints, can be seen as a SNR improvement strategy.
The key premise is that this is a supplementary visual training/aid tool, not a diagnostic or curative device. It's recommended for individuals aged 13 and above, with daily sessions of 30-40 minutes (up to an hour). While it doesn't claim clinical efficacy, it serves as a tool to support daily coordination training. The approach of changing input design to alter perception is not limited to hearing; it's worth considering for other senses as well.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
