PSYC4078A Research Methods in Wits Cognitive Neuroscience Exam Notes

Research Methods in Wits Cognitive Neuroscience (PSYC4078A) examines how cognitive and neural processes are studied with rigorous experimental design, statistical reasoning, and ethical research practice. The course is especially important because cognitive neuroscience depends on methods that can connect behaviour, brain activity, and theory without overclaiming what any single measure can show. Strong exam answers usually demonstrate not only definitions, but also the ability to compare methods, identify threats to validity, and justify why a particular design is appropriate for a given research question.

1. The Logic of Research in Cognitive Neuroscience

Cognitive neuroscience asks how mental functions such as attention, memory, language, perception, decision-making, and executive control are implemented in the brain. Research methods are the tools that make those questions answerable, and the quality of those tools determines the quality of the conclusions. A useful way to approach PSYC4078A is to treat methods not as a list of techniques, but as a set of logical decisions about what counts as evidence.

1.1 What makes cognitive neuroscience different from general psychology?

The central difference is the attempt to relate behavioural performance to neural mechanisms. A standard psychology study may ask whether a distraction affects reaction time. A cognitive neuroscience study asks the same question but adds a brain-based dimension: Which neural systems change under distraction? Is the change local or distributed? Does it occur before the behavioural effect, at the same time, or after it?

This creates a methodological challenge. Brain data are often indirect, noisy, and temporally or spatially limited. Behavioural data are often clean but psychologically underdetermined. A slower reaction time could reflect attention, motivation, fatigue, confusion, motor preparation, or strategic caution. A good method must therefore separate the construct of interest from alternative explanations.

1.2 Core scientific logic: variables, manipulation, and inference

Research methods in this field rest on a few basic distinctions:

Independent variable (IV): the factor manipulated or compared.
Dependent variable (DV): the outcome measured.
Control variables: factors held constant to reduce confounding.
Confounds: alternative explanations that vary with the IV.
Operational definition: the way a theoretical construct is measured in practice.

For example, if a study investigates working memory, the theory may concern capacity or maintenance of information. Operationally, this might be measured through accuracy on a digit span task, load effects on n-back performance, or neural activation in prefrontal and parietal regions during memory maintenance. If the researcher does not clearly define the construct, the findings cannot be interpreted cleanly.

A major exam skill is to explain why correlation is not causation. If a brain region is more active during a task, that does not prove the region causes the behaviour. The region may support the process, reflect a consequence of it, or simply co-activate because of a third variable. Causal claims require stronger designs, such as lesion studies, stimulation methods, or carefully controlled experiments.

1.3 Typical research questions in Wits cognitive neuroscience

The kinds of questions usually addressed in this area include:

How does attention modulate sensory processing?
Which brain systems support memory encoding and retrieval?
What neural changes accompany bilingual language control?
How are executive functions affected by stress, fatigue, or neurological injury?
Can differences in brain activity predict individual differences in performance?

Each of these questions can be studied with several methods, and each method gives a different kind of answer. Behavioural experiments show what people do. EEG and MEG show when processes unfold. fMRI shows where blood-flow changes occur. Lesions show what happens when tissue is damaged. Eye tracking shows gaze allocation. Neuropsychological tests show cognitive profiles. No single method is sufficient on its own.

1.4 The role of theory in method selection

Good research design begins with theory. A method is appropriate only if it matches the hypothesis. For instance:

If the question is about timing, EEG is often better than fMRI because EEG has millisecond temporal resolution.
If the question is about deep brain structures, fMRI may be more suitable than EEG because it has better spatial coverage.
If the question is about causal necessity, lesion studies or brain stimulation may be more informative than purely observational methods.
If the question is about real-time interaction with stimuli, eye tracking or psychophysics may be more suitable.

Theory determines what counts as evidence. A weak design often uses the wrong measure for the question. For example, using only reaction time to infer frontal-lobe executive control may be too indirect if the task also depends heavily on motor speed or visual search. Similarly, inferring “memory” from a single accuracy score can be misleading if the score is influenced by guessing or response bias.

1.5 Levels of explanation

Cognitive neuroscience often works across several levels at once:

Computational level – What problem is the system solving?
Algorithmic level – What representations and operations are used?
Implementation level – How is the process realised in neural tissue?

A useful study may link all three levels, but many studies only address one or two. For example, a Stroop task can reveal conflict effects at the behavioural level, while fMRI can localise associated activation, and computational models can explain how competing representations are resolved. Examiners often reward answers that show awareness that brain localisation alone does not equal explanation. A region may be involved in many tasks, and the same task may depend on several regions working together.

1.6 Why methodological rigour matters

Cognitive neuroscience findings are especially vulnerable to overinterpretation because the brain is complex and the public is eager for simple explanations. Claims such as “this area is the seat of memory” or “this scan shows intelligence” are far too simplistic. Rigorous methods protect against these mistakes by requiring:

Clear hypotheses
Appropriate controls
Sufficient sample sizes
Valid measures
Transparent analysis
Cautious interpretation

In exam writing, the safest and strongest phrasing is often: “The method provides evidence consistent with…” rather than “The method proves…”. That phrasing reflects scientific humility and methodological awareness.

2. Research Designs, Sampling, and Validity

A research method is only as good as the design surrounding it. In PSYC4078A, design questions are central because the same measurement technique can produce excellent or poor data depending on how the study is structured. A high-quality answer should distinguish between exploratory and confirmatory work, experimental and non-experimental designs, and internal, external, construct, and ecological validity.

2.1 Experimental, quasi-experimental, and correlational designs

Experimental designs

In an experiment, the researcher manipulates the IV and randomly assigns participants to conditions. This is the strongest design for causal inference because randomisation helps equalise known and unknown confounds across groups.

Examples in cognitive neuroscience include:

Comparing attention under high-load and low-load conditions
Manipulating sleep deprivation before a memory task
Testing stimulus type in a language comprehension study

The key advantage is causal inference. The limitation is that experiments may be artificial or ethically constrained.

Quasi-experimental designs

Quasi-experiments compare groups that already exist, such as patients with brain injury versus healthy controls, or bilingual versus monolingual speakers. These designs are common in cognitive neuroscience because many questions cannot ethically or practically be manipulated.

The main weakness is selection bias. If groups differ in education, medication, injury severity, or socioeconomic background, group differences may not reflect the variable of interest.

Correlational designs

Correlational studies measure the association between variables without manipulation. For example, one may examine whether working memory capacity predicts academic performance or whether cortical thickness correlates with task accuracy.

These studies are useful for identifying patterns and generating hypotheses, but they cannot establish causation. A correlation may result from a third variable or bidirectional influence.

2.2 Within-subjects and between-subjects designs

A between-subjects design assigns different participants to different conditions. A within-subjects design exposes the same participant to multiple conditions.

Between-subjects advantages

No carryover effects
Simpler to explain
Useful when repeated exposure is impossible

Between-subjects disadvantages

Requires larger samples
More sensitive to individual differences
Lower statistical power if sample size is limited

Within-subjects advantages

Greater statistical efficiency
Each participant serves as their own control
Often higher power with fewer participants

Within-subjects disadvantages

Practice effects
Fatigue effects
Order effects
Demand characteristics

A typical cognitive neuroscience example is a memory experiment where each participant completes both an encoding condition and a control condition while EEG is recorded. This can be powerful, but the order of conditions must be counterbalanced to prevent systematic bias.

2.3 Counterbalancing, randomisation, and control

Randomisation reduces systematic bias in assigning participants or trial orders. Counterbalancing ensures that order effects are distributed evenly across conditions. In repeated-measures studies, counterbalancing is essential when earlier trials may alter performance on later trials.

A simple example:

Half the participants do Condition A then Condition B.
Half do Condition B then Condition A.

If performance differs due to order, counterbalancing helps identify and reduce that problem.

Controls may include:

Standardised instructions
Equal stimulus presentation duration
Similar difficulty across conditions
Blinding of experimenters or participants where possible
Rest periods in long neuroimaging sessions

Control is especially important in cognitive neuroscience because many tasks are affected by alertness, strategy, and fatigue. A well-controlled design can eliminate rival explanations that would otherwise undermine interpretation.

2.4 Sampling and participant considerations

Sampling shapes the kind of conclusion a study can support. A sample of undergraduate volunteers can be convenient, but it may not generalise to older adults, clinical populations, or culturally diverse groups. In South African contexts, this is especially important because language background, educational access, and socioeconomic factors can affect cognitive performance and testing conditions.

Important sampling issues include:

Sample size: too small and results become unstable.
Representativeness: whether the sample reflects the target population.
Attrition: participants who drop out may differ systematically from those who remain.
Inclusion/exclusion criteria: necessary for data quality but may limit generalisability.

In neuroimaging studies, sample sizes are often modest because of cost and time, but small samples increase the risk of false positives and overestimated effect sizes. A good exam answer should note that a “significant” result is not automatically a reliable one if the sample is tiny and the analysis was not planned carefully.

2.5 Validity: four major types

Internal validity

Internal validity asks whether the IV truly caused the DV. Threats include confounds, order effects, experimenter bias, and participant expectations.

External validity

External validity concerns generalisability to other people, settings, tasks, and times. A task performed in a scanner may not reflect everyday cognition because the environment is artificial.

Construct validity

Construct validity asks whether the study really measures the intended concept. For example, does a “working memory” task actually measure working memory, or does it mainly measure processing speed and test familiarity?

Ecological validity

Ecological validity refers to the extent to which the task resembles real-world cognition. A highly controlled laboratory task may sacrifice realism for precision.

These forms of validity often trade off against one another. A highly controlled experiment may have strong internal validity but weak ecological validity. A field-based study may look realistic but be harder to control. The best design depends on the research question.

2.6 Ethics and participant welfare in design

Ethics is not a separate topic from methods; it shapes the design itself. Cognitive neuroscience studies may involve:

Fatigue from long testing sessions
Anxiety in scanners
Discomfort from electrodes or stimulation
Confidentiality concerns about brain and health data
Special protections for clinical populations or minors

Ethical design requires informed consent, the right to withdraw, minimising harm, and appropriate debriefing. If a participant experiences distress in a task involving emotional images or traumatic content, the study must have a response plan. Good methodology protects people as well as data quality.

3. Behavioural Methods and Task-Based Measurement

Behavioural methods are the foundation of cognitive neuroscience because brain measures must be interpreted in relation to performance. Without behavioural data, neural activation is difficult to understand; without neural data, behavioural outcomes may remain theoretically broad. PSYC4078A typically expects students to know how core cognitive tasks work, what they measure, and what their limitations are.

3.1 Reaction time and accuracy

The most common behavioural measures are reaction time and accuracy. Reaction time is sensitive to processing difficulty and decision demands, while accuracy reflects success or error in the task. However, each measure alone can be misleading.

A participant may be fast but inaccurate, suggesting speed–accuracy trade-offs. Another may be slow but accurate, suggesting cautious response strategies. Therefore, both measures should usually be analysed together.

Common interpretations:

Slower reaction time may indicate greater processing load.
Lower accuracy may indicate poorer discrimination, weaker memory, or attentional lapses.
A change in both can suggest true performance impairment.
A change in one but not the other may suggest strategy rather than capacity.

3.2 Classic cognitive tasks and what they measure

Stroop task

The Stroop task measures interference control and selective attention. Participants name the ink colour of a word that may be congruent or incongruent with the word’s meaning. Incongruent trials usually produce slower and less accurate responses.

The task is useful because it reveals conflict between automatic reading and controlled colour naming. However, Stroop performance can be influenced by language proficiency, reading speed, and familiarity with the stimuli.

Go/No-Go task

This task measures response inhibition. Participants respond to frequent “Go” stimuli and withhold responses to rare “No-Go” stimuli. It is often used to examine impulsivity, executive control, and prefrontal functioning.

The key limitation is that poor performance may reflect attention failure rather than inhibition specifically.

Flanker task

The Flanker task measures selective attention and conflict monitoring. A central target is surrounded by distracting flankers that may point in the same or opposite direction. Incongruent flankers slow responses and increase errors.

N-back task

The n-back task measures working memory updating and monitoring. Participants must decide whether each stimulus matches one presented n items earlier. Higher n increases memory load.

Critics note that n-back tasks may depend heavily on continuous updating, attentional vigilance, and strategy use, so they may not capture all aspects of working memory.

Digit span and span-based measures

Digit span tasks assess short-term and working memory capacity. They are useful but can be limited by language, rehearsal strategies, and familiarity with number sequences.

3.3 Psychophysics and threshold measurement

Psychophysics examines how physical stimulus properties relate to perception. It is essential in cognitive neuroscience because it helps isolate sensory processing from higher cognition.

Typical psychophysical questions include:

What is the smallest intensity difference a person can detect?
How quickly does a stimulus become perceptually identifiable?
How does masking alter conscious perception?
At what point does stimulus frequency appear continuous rather than discrete?

Important concepts include:

Threshold: the level at which a stimulus is detected or discriminated.
Sensitivity: the ability to detect small differences.
Signal detection: distinguishing signal from noise.

Psychophysics is valuable because it gives precise measures of perceptual function. It also helps identify whether apparent “cognitive” effects are actually perceptual. For example, a person who performs poorly on a visual memory task may have a visual discrimination problem rather than a memory problem.

3.4 Signal detection theory

Signal detection theory separates sensitivity from response bias. Sensitivity reflects how well a person can distinguish signal from noise; bias reflects the tendency to say “yes” or “no” when uncertain.

This distinction matters greatly in memory and perception research. A participant may appear to have poor memory because they are conservative and avoid false alarms, not because they cannot recognise targets. Exam responses should emphasise that accuracy alone does not capture both sensitivity and decision strategy.

Common outcomes:

High sensitivity, neutral bias: ideal performance
Low sensitivity, conservative bias: misses many true signals
Low sensitivity, liberal bias: many false alarms

3.5 Eye tracking and overt attention

Eye tracking measures gaze direction, fixation duration, saccades, and scan patterns. It is especially useful for studying attention, reading, visual search, and social cognition.

Why it matters:

Gaze often predicts where processing is taking place.
Fixation duration can reflect difficulty or deeper processing.
Saccade patterns reveal exploration strategies.
Eye movements help distinguish covert from overt attention.

For example, in a reading study, longer fixations on complex words may indicate lexical difficulty. In a visual search task, efficient participants may show shorter search times and more direct gaze paths. In social cognition, gaze toward faces, eyes, or emotionally salient features can indicate processing priorities.

3.6 Behavioural data quality

Behavioural data are affected by:

Poor instructions
Equipment timing errors
Uncontrolled distractions
Fatigue
Learning effects
Guessing and lapses of attention

Researchers often clean data by excluding implausible reaction times, removing practice trials, and checking for outliers. However, exclusion rules must be justified in advance or applied consistently. Arbitrary data trimming can distort findings. An excellent exam answer can mention that the same behavioural dataset may yield different conclusions depending on how outliers are handled, which is why transparency is crucial.

3.7 Why behavioural methods remain indispensable

Even when advanced neuroimaging is available, behavioural methods remain indispensable for three reasons:

They provide the functional outcome that brain data must explain.
They are often more reliable and easier to interpret than neural signals.
They can reveal whether a neural change has real performance consequences.

A scan can show activation, but only the behavioural result shows whether the participant actually succeeded, failed, guessed, or changed strategy. This is why strong cognitive neuroscience studies almost always integrate performance measures with neural measures rather than treating them separately.

4. Neuroimaging and Brain-Based Measurement Methods

The hallmark of cognitive neuroscience is the use of methods that connect cognition with neural activity. Each technique has a particular trade-off between spatial resolution, temporal resolution, invasiveness, cost, and interpretability. PSYC4078A requires a clear understanding of what each method can and cannot tell us.

4.1 EEG and ERP

What EEG measures

Electroencephalography (EEG) records electrical activity from the scalp produced by synchronised postsynaptic potentials in populations of neurons. It is especially good for tracking rapid neural changes.

Strengths of EEG

Excellent temporal resolution
Non-invasive
Relatively affordable compared to many imaging methods
Useful for studying attention, perception, and timing of cognitive processes

Weaknesses of EEG

Poor spatial localisation
Signal contamination from muscle movement, eye blinks, and electrical noise
Complex inverse problem: scalp signals do not map uniquely to sources
Sensitive to participant movement and preparation quality

Event-related potentials

ERPs are time-locked EEG responses to specific events. They are extracted by averaging across multiple trials to improve the signal-to-noise ratio.

Common ERP interpretations:

Early components often reflect sensory processing.
Later components may reflect attention, categorisation, or memory updating.

A strong exam answer should stress that ERP components are not “modules” in a simplistic sense. Their meaning depends on the task, timing, and experimental context. The same component may reflect different processes across studies.

4.2 fMRI

What fMRI measures

Functional magnetic resonance imaging (fMRI) measures changes in blood oxygenation, usually through the blood-oxygen-level-dependent signal. It is an indirect measure of neural activity, based on the relationship between neural firing and local haemodynamic response.

Strengths of fMRI

Good spatial resolution
Whole-brain coverage
Useful for mapping distributed networks
Suitable for many cognitive domains

Weaknesses of fMRI

Poor temporal resolution compared with EEG
Indirect measure of neural activity
Expensive
Sensitive to motion artefacts
The BOLD signal reflects vascular changes, not neurons directly

A common misconception is that fMRI “shows the brain thinking.” More accurately, it shows where blood-oxygen changes correlate with a task. That distinction matters because vascular and neural signals are related but not identical.

4.3 MEG

Magnetoencephalography (MEG) measures magnetic fields generated by neural activity. Like EEG, it offers excellent temporal resolution, but it often has better source localisation than EEG under the right conditions. It is less common in many teaching settings because it is very expensive and requires specialised infrastructure.

MEG is especially useful for:

Auditory processing
Language timing
Sensorimotor integration
Rapid oscillatory dynamics

Its major limitation is access and cost, which means many students will encounter it more in theory than in practice.

4.4 Lesion studies and neuropsychology

Lesion methods study the effects of brain damage on cognition. They are powerful because they can support causal claims: if damage to a region is associated with a specific deficit, that region may be necessary for that function.

Examples include:

Aphasia after left-hemisphere language damage
Memory impairment after medial temporal damage
Spatial neglect after right parietal damage

However, lesion studies are complicated by:

Variability in lesion size and location
Brain plasticity and compensation
Pre-existing differences among patients
Co-occurring impairments affecting performance

A lesion does not simply “erase” a function. It may disrupt a network, alter connectivity, or force compensatory strategies. The most accurate interpretation usually involves networks rather than isolated regions.

4.5 Brain stimulation methods

TMS

Transcranial magnetic stimulation uses magnetic pulses to modulate cortical activity. It can temporarily disrupt or facilitate processing in targeted areas, making it valuable for causal inference.

Advantages:

Can test necessity of a region
Non-invasive
Temporally precise compared with lesion studies

Limitations:

Limited to accessible cortical regions
Effects can be variable
Not all tasks are suitable
Safety and comfort considerations are important

tDCS

Transcranial direct current stimulation applies weak electrical currents to alter cortical excitability. It is often used in cognitive enhancement or rehabilitation research.

Advantages:

Relatively simple and low-cost
Portable
Can be combined with behavioural tasks

Limitations:

Effects are subtle and variable
Spatial specificity is low
Mechanisms are still debated
Placebo control is important

4.6 Comparing methods: a practical overview

Method	Main measure	Best for	Strength	Limitation
EEG	Electrical activity at scalp	Timing of cognition	High temporal resolution	Poor spatial localisation
ERP	Time-locked EEG response	Event-related processing	Fine-grained timing	Requires many trials
fMRI	BOLD signal	Spatial localisation and networks	Good spatial coverage	Indirect and slow
MEG	Magnetic fields	Fast neural dynamics	High temporal resolution and useful localisation	Expensive, limited access
Lesion studies	Cognitive deficits after damage	Causal necessity	Strong causal inference	Rare, variable lesions
TMS	Temporary cortical disruption	Testing necessity of cortex	Causal and time-sensitive	Limited depth, variable effects
tDCS	Modulated excitability	Intervention and facilitation	Cheap and portable	Small, inconsistent effects

4.7 Choosing the right method

Method choice should be driven by the hypothesis. Consider three example questions:

When does a visual prediction error occur?
EEG or MEG is ideal because timing is crucial.
Which network supports episodic memory retrieval?
fMRI is suitable because spatial localisation and network mapping are needed.
Is a frontal region necessary for inhibition?
TMS or lesion evidence is stronger than correlation alone.

A sophisticated answer can also note that many studies combine methods. For example, EEG may be used to identify timing and fMRI to identify localisation. This multimethod approach improves confidence because each technique compensates for the weaknesses of the other.

4.8 Interpretation and caution

Brain measures should be interpreted conservatively. Activation does not necessarily mean importance, and lack of activation does not necessarily mean absence of processing. A region may show no effect because of sensitivity limits, task design, or analytic choices. Likewise, activation could reflect effort, arousal, error monitoring, or general task demands rather than the target construct itself.

In exam language, the safest formulation is:

“This method suggests involvement of…”
“The finding is consistent with…”
“Causal inference is limited unless…”

That caution is not weakness; it is methodological maturity.

5. Data Analysis, Ethics, and Exam Strategy

The final major component of PSYC4078A is understanding how research findings are evaluated, analysed, and communicated responsibly. Cognitive neuroscience generates complex datasets, but analysis choices are themselves methodological decisions. Ethical practice and sound interpretation are therefore inseparable from the technical methods.

5.1 Descriptive statistics and inference

Most studies begin with descriptive statistics:

Mean
Median
Standard deviation
Range
Variance
Error bars and confidence intervals

These summaries help describe central tendency and variability. But descriptive statistics alone do not answer whether a difference is likely to reflect a real effect rather than random variation.

Inferential statistics allow researchers to test hypotheses. Common approaches include:

t-tests for comparing two means
ANOVA for comparing multiple conditions
Correlation for associations
Regression for prediction
Mixed models for repeated measures and hierarchical data
Non-parametric tests when assumptions are violated

In cognitive neuroscience, analysis often becomes more complex because data are nested: trials within participants, participants within groups, and regions within networks. Statistical models must respect that structure.

5.2 Reliability, power, and effect sizes

A result is only useful if it is reliable. Reliability refers to consistency across repeated measurement. A task with poor reliability can produce unstable findings, even if the average effect looks impressive.

Key concepts:

Statistical power: probability of detecting a true effect.
Effect size: magnitude of the effect, not just whether it is statistically significant.
Confidence interval: range of plausible values for the estimate.

Small samples are a major weakness in many neuroscience studies because they reduce power and inflate the risk of false positives. An underpowered study may still produce a significant p-value, but the estimate is often unstable and difficult to replicate.

For exam purposes, it is helpful to remember:

Large samples improve precision.
Reliable tasks improve confidence in individual differences.
Replication strengthens the credibility of a finding.

5.3 Multiple comparisons and false positives

Brain data often involve many tests at once. For example, voxel-wise fMRI analysis may test thousands of locations, and EEG may test many time points and electrode sites. If each test is treated separately at the same significance threshold, the chance of false positives increases dramatically.

Methods used to address this include:

Bonferroni correction
False discovery rate control
Cluster-based correction
Pre-registered analysis plans

The core idea is simple: more testing means more opportunities for chance findings. Good methodology must control this inflation or interpret results cautiously.

5.4 Open science and reproducibility

Reproducibility has become central in cognitive neuroscience because analytic flexibility can unintentionally produce misleading results. Open science practices improve trustworthiness by making research more transparent.

Important practices include:

Pre-registration of hypotheses and analysis plans
Sharing code and data where ethical and legal constraints allow
Reporting all conditions and exclusions
Distinguishing exploratory from confirmatory analyses
Replicating key findings across samples

A study is more convincing when its conclusions can be reproduced with similar results by independent researchers. This is especially important in a field where multiple preprocessing pipelines and statistical choices can influence outcomes.

5.5 Ethics in data handling and reporting

Ethics extends beyond participant protection to include honest reporting. Researchers must avoid:

Selective reporting of only significant results
HARKing, or hypothesising after results are known
p-hacking through repeated unplanned analyses
Misrepresentation of weak or ambiguous findings

Brain and cognitive data may also be sensitive. Neuroimaging datasets can reveal health information, structural differences, or incidental findings. Confidentiality and responsible storage are therefore essential.

In addition, cognitive neuroscience in South Africa should be attentive to contextual fairness. Measures developed in one language or cultural setting may not transfer neatly to another. Testing materials, instructions, and norms must be adapted carefully to avoid bias. A method can be technically valid yet socially unfair if it systematically disadvantages certain groups.

5.6 Common exam pitfalls

Students often lose marks by doing one or more of the following:

Confusing correlation with causation
- Example: “This region causes memory because it is active during memory tasks.”
- Better: “The region is associated with memory processing, but causal evidence requires additional methods.”
Overstating localisation
- Example: “This task is controlled by one brain area.”
- Better: “The task likely recruits a distributed network.”
Ignoring behavioural data
- Brain activation without performance can be hard to interpret.
Using broad terms without definition
- Terms like attention, executive function, and working memory must be specified.
Not discussing limitations
- Every method has trade-offs; acknowledging them strengthens the answer.
Forgetting ethics
- In human research, ethics is not optional.

5.7 How to answer method questions well

A strong exam response usually follows a pattern:

Define the method or design clearly
Explain what it measures
State why it is appropriate for the question
Identify strengths
Identify limitations
Suggest improvements or complementary methods

For example, if asked about EEG in a study of attention, a good answer may say that EEG is ideal for capturing the millisecond timing of attentional shifts, but its spatial precision is limited. It would then mention that combining EEG with fMRI or behavioural data could improve interpretation.

5.8 Integrating the whole course

The most important insight in PSYC4078A is that research methods are not separate from theory; they are the means by which theory is tested. Cognitive neuroscience advances when researchers choose methods that fit the question, control their confounds, analyse their data responsibly, and interpret findings with restraint.

A high-quality understanding of this course includes the following integrated principles:

Behavioural measures establish what participants did.
Brain measures suggest where and when processing occurred.
Experimental design determines whether causal inference is possible.
Statistics determine whether the result is robust or likely to be noise.
Ethics governs whether the study is acceptable and responsible.
Interpretation determines whether the finding genuinely advances knowledge.

In revision, it helps to think in terms of method combinations. A memory study might use a within-subjects design, reaction time and accuracy measures, EEG for timing, and strict counterbalancing to control practice effects. A clinical study might use neuropsychological tests, lesion evidence, and group comparisons. A language study might pair eye tracking with fMRI to understand both attention and neural localisation. These are not separate topics; they are different ways of answering the same scientific question.

Ultimately, the best exam answers in Research Methods in Wits Cognitive Neuroscience show that the student understands the limits of evidence as well as its power. That balanced judgement is the hallmark of a competent cognitive neuroscientist: precise about what a method can show, careful about what it cannot, and thoughtful about how multiple methods together produce stronger scientific conclusions.