The associate type guide

What is the meaning of the data in each associate type?

This is a good question that most people who are new to the software ask. The answer is that there is no super-easy way to figure out what each associate type is, although we are striving to make this easier. Since I (SDV) spent many years primarily coding for myself, I did not document these types very well. However, I will try to improve this going forward, by using the following guidelines:

• 1. the "owner" should be the name of the .m file that produces the associate; this will allow one to read the code that generated the associate

  2. the "desc" or description field should say a lot about what the measure is

Nevertheless, there are several rules of thumb for interpreting the associate types.

Associate type name rules of thumb

1. '<stimulus> test' and '<stimulus> resp'

The directory where most stimuli are run are coded as '<stimulus> test', and the average responses that were recorded to each stimulus presentation are coded as '<stimulus> resp'.

For example, 'FineDir test' is commonly used to indicate the directory where data from the "fine direction test" (that is, the direction test run with the finest steps in direction angle and run at the best spatial and temporal frequency), and 'FineDir resp' would contain the stimulus angles that were tested, the average responses and standard deviations and standard errors:

B1 = findassociate(celldata{i},'FineDir test','','')

should give

B1 =

type: 'FineDir test'

owner: 'gratingsheetlet'

data: 't00039'

desc: 'FineDir test'

and

B2 = findassociate(celldata{i},'FineDir resp','','')

should give

B2 =

type: 'FineDir resp'

owner: 'gratingsheetlet'

data: [1x1 struct]

desc: ''

2. Type names like 'SP F0 Ach OT Response curve' or 'TP Ach OT Response curve'

Since about 2007, we have made an effort to standardize some of our associate type names so that responses that are recorded via fundamentally different methods (electrical recordings vs. calcium imaging, for example) or differ only in color (achromatic vs. S-cone driving or ML-cone driving, for example) can use many of the same analysis programs. When one looks at a list of associates, one may see several items with 'SP F0 Ach OT' in front, or 'TP Ach OT'.

'SP F0 Ach OT Response curve' means this is spiking data ('SP'), it is the F0 component (or the "mean" response, as opposed to a modulated response), the stimulus was achromatic 'Ach', and that it is an orientation tuning curve result ('OT').

The prefixes that are standardized are as follows:

• • 'SP F0' -- spike data, F0 ("mean" response)

  • 'SP F1' -- spike data, F1 (the response that is modulated at the stimulus temporal frequency)

  • 'SP F2' -- spike data, F2 (the response that is modulated at twice the stimulus temporal frequency)

  • 'TP' -- Two-photon data, in change in fluorescence (that is, ∆F/F)

The colors that are standardized are as follows:

• • 'Ach' -- Achromatic (black and white)

  • 'S' -- S-cone-isolating stimulus (tree shrew)

  • 'ML' -- ML-cone-isolating stimulus (tree shrew)

The tuning curve types that are standardized are as follows:

• • 'OT' -- orientation tuning (includes direction tuning)

  • 'CT' -- contrast tuning

  • 'SF' -- spatial frequency tuning

  • 'TF' -- temporal frequency tuning

  • 'CEDE' -- color "Dacey" extended, as in Figure 5, Johnson/Van Hooser and Fitzpatrick 2010

For a big list and a description of associates that are calculated by the standard tuning curve analysis functions, see

• help otanalysis_compute

• help ctanalysis_compute

• help sfanalysis_compute

• help tfanalysis_compute

Orientation/direction analysis:

The full compliment of associates computed for orientation and direction analysis can be seen at help otanalysis_compute. These are the most common types:

• • 'OT vec varies p' : Hotelling's T^2 test of orientation vector data, to see if there is any significant orientation tuning to speak of

  • 'OT visual response p' Anova p across stims and blank (if available) (answers if cell is visually responsive)

  • 'OT Circular variance' - circular variance in orientation space

  • 'OT Dir Circular variance' - circular variance in direction space

  • 'OT Fit Pref' Direction preference from double gaussian fit (may be garbage if 'OT vec varies p' is not significant)

  • 'OT Fit Orientation index blr' - Orientation index w/ blank subtracted but rectified so blank <= Orth (may be garbage if 'OT vec varies p' is not significant)

  • 'OT Fit Direction index blr' Direction index w/ blank subtracted but rectified so blank <= Rn (may be garbage if 'OT vec varies p' is not significant)

  • 'OT Tuning width' Tuning width (half width at half height) (may be garbage if 'OT vec varies p' is not significant)