Experiment framework that provides rapid experiment GUI design and prototyping using Enthought's Traits and TraitsUI system. With a few lines of code, you can define the parameters for your experiment and have them logged instantly to a table in a HDF5 file as you run through each trial. The framework supports:

• Specifying parameters as an expression, allowing for:
  • Dynamically computing the value of a parameter as a function of one or more other parameters.
  • Roving a parameter from trial to trial.
  • Using generators to provide a pre-determined parameter order (e.g. ascending, descending and/or blocked sequences).
• The value of parameters on each trial are automatically logged to a HDF5 table. Each row contains the value of all parameters for that given trial and the value of the expression that generated the parameter value (e.g. if you have a parameter called poke_duration that is set to random(0.2, 0.4), it will save both the computed value as well as the expression string.
• Parameter expressions can be updated on-the-fly via the GUI (using an apply/revert button to ensure that changes are registered at the appropriate time).
• Built-in safety checks to ensure that you do not accidentally enter an invalid expression or circular reference in which the values of two parameters depend on each other.

To use this framework you need to inherit from the experiment.AbstractController class and provide logic that is specific to your experiment. The superclass provides logic for handling changes to values of the parameters on each trial. These changes occur when the user modifies the value in the GUI and hits the apply button or the value of the parameter is specified by an expression whose output changse from trial to trial.

When the value changes, the controller needs to know how to handle this change. For example, if you are using a infusion pump whose rate is programmable, then you probably need to send a command via the serial port to update the pump rate setting. To provide this logic, define a method called set_parameter_name that takes a single argument (the updated value).

For example:

def set_pump_rate(self, value):
    self.serial.writeln('RATE {:.4f}'.format(value))

At the beginning of every trial refresh_context should be called to set up the paradigm variables. Since the evaluation of expressions may depend, in part, on values you provide via set_current_value, expressions are not evaluated until you call get_current_value. At some point before starting the trial you should be sure to call evaluate_pending_expressions to be sure the remaining values get computed.