Table 1:

Multi-information I estimated either directly from the data or from maximum entropy models, for random subpopulations of 10 and 20 neurons (100 subpopulations each), as well as the ratio of the multi-information between model and data

		Data	Minimal	Linear	Complete
N = 10	I	0.0713 ± 0.0398	0.0343 ± 0.0268	0.0478 ± 0.0315	0.0498 ± 0.0326
		1	0.444 ± 0.135	0.649 ± 0.108	0.677 ± 0.108
N = 20	I	0.3188 ± 0.0967	0.1291 ± 0.0558	0.1702 ± 0.0592	0.1789 ± 0.0606
		1	0.393 ± 0.071	0.531 ± 0.055	0.557 ± 0.055

Results are reported as the Mean (± SD) in bits.