# Dataset Structure

The decoding scans are provided as .h5 files. There are two types of decoding scans:
- *Objects*: For the object discrimation task
- *Ellipse*: For the detection task

The following attributes are present in each individual dataset:
- **pupil_trials:** mean pupil size for each trial
  - shape:`n_trials,`
- **class_trials:** image class for each trial. See below for details.
  - shape:`n_trials,`
- **activity_trials:** mean activity per neuron for each trial 
  - shape:`n_neurons x n_trials`



## `class_trials` description
### Object discrimination task

##### Paradigm 1
- `o1BLfgv2`: object 1, green channel, foreground (Fig. 5)
- `o2BLfgv2`: object 2, green channel, foreground (Fig. 5)
- `o1UVfgv2`: object 1, UV channel, foreground (Fig. 5)
- `o2UVfgv2`: object 2, UV channel, foreground (Fig. 5)

##### Paradigm 2
- `o1BLfbv2`: object 1, green channel, foreground & background & positive contrast (Extended Data Fig. 10)
- `o2BLfbv2`: object 2, green channel, foreground & background & positive contrast (Extended Data Fig. 10)
- `o1UVfbv2`: object 1, UV channel, foreground & background & positive contrast (Extended Data Fig. 10)
- `o2UVfbv2`: object 2, UV channel, foreground & background & positive contrast (Extended Data Fig. 10)

##### Paradigm 3
- `o1BLfdv2`: object 1, green channel, foreground & background & negative contrast (Extended Data Fig. 10)
- `o2BLfdv2`: object 2, green channel, foreground & background & negative contrast (Extended Data Fig. 10)
- `o1UVfdv2`: object 1, UV channel, foreground & background & negative contrast (Extended Data Fig. 10)
- `o2UVfdv2`: object 2, UV channel, foreground & background & negative contrast (Extended Data Fig. 10)

##### Paradigm 4
- `o1BLnbv2`: object 1, green channel, no background/foreground (Extended Data Fig. 10)
- `o2BLnbv2`: object 2, green channel, no background/foreground (Extended Data Fig. 10)
- `o1UVnbv2`: object 1, UV channel, no background/foreground (Extended Data Fig. 10)
- `o2UVnbv2`: object 2, UV channel, no background/foreground (Extended Data Fig. 10)

### Detection task
- `green`: dark ellipse is present in the green color channel
- `uv`: dark ellipse is present in the UV color channel
- `noise`: no ellipse is present