Probing neural signals – with stimuli, tasks and populations
Each neuroimaging study requires a tailored experimental design that allows questions about brain anatomy or function to be answered in a very controlled way.
When the question concerns the structure of the brain, a typical experiment will compare how anatomical brain scans differ in different populations of participants, or in the same participant scanned at multiple different time points. When different populations are being compared, the experimental design aims to match these populations as closely as possible for factors that are not of interest (e.g. age, gender) so that differences in brain structure between populations reflect the factor of interest (e.g. type of symptoms).
When the experiment aims to find out which parts of the brain are involved in a function of interest (e.g. speaking), we need to measure where in the brain activity increases when participants are engaged in tasks that require this function (activation tasks) compared to when they are engaged in tasks that do not require that function (baseline tasks). As the activation task typically involves multiple functions, the greatest challenge is to identify suitable baseline tasks that control for all the functions of no-interest in the activation task.
To address this challenge, studies of behaviour have revealed a number of rules that determine a good experimental design for functional imaging experiments. For example:
- Factorial designs ensure that different functions are manipulated independently of one another;
- Parametric designs correlate the degree to which a function is thought to be engaged with brain activity;
- Multiple regression allows the influence of variables of no interest to be factored out;
- Counterbalancing ensures that the order of conditions cannot explain the measured brain responses of interest.
Designing a task-based neuroimaging experiment also requires us to select stimuli that are controlled across conditions, and decide how frequently the stimuli are presented and how long they will be presented for. All these decisions are critically dependent on the type of functional imaging experiment that is being conducted.
For example, for fMRI experiments:
- The optimal time that participants should be engaged in one task before switching to another depends on the timing of the haemodynamic response that is being measured (15-30seconds);
- Stronger responses are measured when the stimuli are closer together because the response to each stimulus summates into a bigger response; and
- Different parts of the brain (slices) are measured at different times, therefore the timing of stimuli needs to ensure that data are acquired from each slice at multiple different time points.
To summarise, designing functional imaging experiments involves:
- Finding behavioural tasks that do and do not engage functions of interest
- Imaging the brain while participants carry out the tasks
- Measuring how quickly and accurately the participants perform the tasks
- Observing which regions of the brain are more involved in the tasks of interest compared to the baseline tasks.
When we learn which brain regions respond to specific tasks, we also know which tasks are needed to activate regions of interest. These tasks can be used by surgeons to test the response in and around targeted regions so that they can avoid damaging parts of the brain that are essential for well-being.
Statistical Parametric Mapping
Our leading software for analysing neuroimaging data was developed at FIL.