It's a good way of understanding spatial streams. Each pixel is a highly-directional transmit antenna, and each of your rods and cones is a highly-directional receiver. The result is 1920x1080 streams that you can interpret all at once!

(Of course, given the extremely wide bandwidth of visible light and the high SNR between your TV and other light sources, you don't need that many spatial streams to beam the information representing HDTV across space. But humans.)

MIMO doesn't operate via "spatial" streams, the antennas are not "highly directional" to the extent that a tx antenna and an rx antenna get an interference free channel.

True on a very technical level. MIMO is required to implement spatial streams. It's also the foundation for beamforming and diversity coding. But technically, MIMO itself is not beamforming or spatial streams.

So while not a 100% perfect example, I don't think the original comment is all that terrible.