The following techniques measure cognitive load directly rather than relying on a second- hand marker and avoid bias that may be caused by (for example) the subjectivity of participant self-reporting.
Functional Magnetic Resonance Imaging (fMRI)
Magnetic resonance imaging (MRI) is well known for its value in the detection of
structural lesions, but over the last 20 years, technological advancements in the hardware and software have enabled MRI to be successfully employed in the study of physiological processes such as brain function (Faro and Mohamed, 2010). fMRI can combine the morphological data relating to the structure of the brain, with functional data that relate to brain activity.The principle relies on a phenomenon known as nuclear magnetic resonance whereby naturally occurring hydrogen atoms in the human body can be made to absorb and then release energy - generating a detectable signal in the process
metabolises more oxygen when activated than when at rest and allows the researcher to view activation patterns that correspond to cognitive activity. When an area of the brain is in use, the cells burn glucose as part of the metabolic process. This process requires oxygen, as supplied in the haemoglobin of red blood cells. This oxygen is replaced by in influx of fresh oxyhaemoglobin, which provides a subtle intrinsic contrast mechanism that causes a statistically measurable increase in signal from a tissue volume when compared to the same area at rest (Faro and Mohamed, 2010). It has been recently hypothesised that the different regions of the brain are functionally specialised, even at quite a fine level (Rodriquez–Moreno and Hirsh, 2006) and research has shown that functional mapping is possible for various regions of the brain including those involved in cognition. Early fMRI studies demonstrated that the dorsal anterior cingulate cortex and the
dorsolateral prefrontal cortex are involved in cognition and play a fundamental role in the mechanism of working memory (Pochon, et al., 2001). More recent research reviewed by Whelan in 2007 has shown that fMRI may also be sensitive enough to be used to identify the parts of the brain that are affected by intrinsic, extraneous and germane cognitive load.
Although the technique is expensive and can therefore only be used on a small sample, it is thought that some effects shown in a single subject are likely to be generalizable, although there will be some effects that are specific to the individual only. Results that are reproduced in more than one test subject are more likely to be generalizable and current research has suggested that a minimum number of participants should be between 10 – 20 to realise a p-value of =< .05 (Faro and Mohamed, 2010). Participant recruitment can be difficult, as suitable volunteers must satisfy certain rigid conditions. For ethical reasons, it is desirable that they have had a previous MRI study of the brain to rule out the possibility of diagnosing incidental pathology during the research scanning. The fact that MRI provides a direct, objective measurement of cognitive load made it a promising consideration for this study, however, there are also many factors that make its use unfeasible. fMRI ideally requires a magnetic field strength of 3T (teslas) or greater; this is approximately 60,000 times more powerful than the earth's magnetic field. As a result, the procedure can pose potentially serious safety issues to research participants. These include contraindication in early pregnancy, damage to implanted devices such as pacemakers and cochlear implants, torque effects on ferromagnetic foreign bodies or
implanted surgical clips and other considerations such as projectile hazards from
ferromagnetic items. Powerful magnetic attraction would rule out the use of m‑learning devices in the research as these typically contain ferromagnetic components and could become a projectile hazard (Shellock, 2007). Furthermore, MRI examinations deploy pulses of low energy electromagnetic radiation which may under rare circumstances cause burns and damage to implanted devices or the mobile devices being used in the activity. Smartphones and tablets also transmit radio-frequency pulses at frequencies that cause artefactual appearances on MRI images (Westbrook, Kaut-Roth and Talbot, 2011). From a research methodology viewpoint, a potentially confounding factor in using fMRI to measure CL is that the MRI environment is, in itself, very distracting. The scanner has high acoustic noise levels - in excess of 100 decibels, and the volunteer would be required to be confined, lying flat in a cylindrical scanner-bore, with a large detector coil
positioned around the head (Westbrook, Kaut Roth and Talbot, 2011). These conditions make interaction with the mobile device difficult, and they present a very artificial
environment, quite unlike the usual study environment for a student using a touch-screen device. These factors may make the results less generalizable and also affect the
ecological validity of the study. Finally, because MRI poses a risk of discovering incidental pathology in the participant, all of the images taken would require to be reported by a radiologist. Image reporting would have added a great deal of further expense. For these reasons, it was considered impractical and ethically questionable to use fMRI for this study.
Electroencephalography (EEG)
In what is otherwise a diminishing clinical field, the evaluation of neurocognition by EEG has been described by Schomer and Lopes Da Silva (2011) as the most fascinating aspect of modern practice in this field, largely because of recent technological developments that allow EEG to compete with fMRI and other more complex techniques such as single photon emission tomography (SPECT). Antonenko, et al. (2010) concur with this view stating that being able to spot subtle fluctuations in CL with a high degree of temporal resolution (i.e. the ability to witness effects in real-time) can help to explain the effects of interventions that could not be assessed using self-evaluation post-testing. However, EEG
measurement can be a time-consuming and inconvenient procedure. When undergoing tradition clinical EEG techniques, the subject is typically required to wear 20 (or more) electrodes, which adhere to the scalp with a conductive gel. Application and removal of the electrodes is a time-consuming process, and requires preparation by the participant (washing the hair before and after the test). Multiple-electrode EEG is therefore
impractical for a large sample of participants. In 2010, Haapalainen, Kim and Dey
discovered that it was possible to make EEG recordings using a single dry electrode worn as a headset. The sensor used on such devices makes use of a new technology that uses noise cancellation, digital filtering and amplification to provide high-quality, research- standard EEG without the need for multiple wet-conductive electrodes. Single-electrode EEG is a quicker, less obtrusive technique, and as the electrode is connected via the Bluetooth wireless protocol, there are no wired connections. The above considerations combined with the relatively low cost of the device and minimal safety hazards made EEG an appealing data collection method for measuring cognitive load, where obtrusive external influences may affect the validity of the study. A dry-electrode EEG set was purchased for testing, but it was discovered that sensitivity of the single electrode did not appear to be high enough for the requirements of the test. This technique requires individual testing of participants and it would have been impractical to test the number required to achieve statistical power in the time available. For these reasons EEG was rejected as a data collection method for this study.
Dual-Task Technique
This method requires the participant to perform a secondary task (such as regular finger tapping) simultaneously with the learning task. The performance in the secondary task being a measure of how much load is being placed on the individual by the primary task. (typically an increase in reaction time is seen in the secondary task with increasing primary task load) (Brünken, Plaas and Leutner, 2003).
Drawbacks to this method include the necessity to be able to measure reaction time accurately and the fact that finger tapping (for example) is obtrusive to the primary task. Brünken, Plaas and Leutner identify this as a limitation stating that even a simple
secondary task may affect the learning outcomes of the primary task. This technique also requires individual testing of participants, and it would have been impractical to test the
number of participants needed to achieve statistical power in the time available for the study. For these reasons, this method was discounted.
Near Infra-Red Spectroscopy
This technique uses optodes (“optical electrodes”) to record the haemodynamic activity in the pre-frontal cortex of the brain relating to oxygenation levels. Near infra-red spectroscopy is a safe, non-invasive technique that shows cognitive load in real time and can be used in any environment. It also shows the areas of the brain cortex that are in use (Durantin, et al., 2014). It was not considered suitable for this study, as the equipment is highly specialised, expensive to hire, and requires a high level of user training. This technique also requires individual testing of participants and it would have been
impractical to test the number of participants required to achieve statistical power in the time available for this study.