H567/02 Section A - Maguire et al. (2000): taxi drivers

Background, design and sample

The hippocampus is involved in memory, including spatial representation and navigation. Animal research had suggested that demanding spatial experience could be related to hippocampal organisation. Licensed London taxi drivers offered an adult human group with extensive navigational experience: learning “the Knowledge” requires detailed familiarity with London's routes, rather than simply following the same journey each day.

The aim was to investigate whether regional hippocampal structure differed between taxi drivers and controls and whether it related to navigation experience. The main comparison was quasi-experimental, using an existing occupational group. Researchers did not randomly assign people to become taxi drivers. A separate correlational analysis related experience duration to hippocampal measures. Calling the scanner session a laboratory procedure does not make taxi experience an experimentally manipulated independent variable.

The driver sample comprised 16 healthy, right-handed male licensed London taxi drivers, aged 32–62, with mean age 44. They had been licensed for at least 1.5 years; licensed experience averaged 14.3 years and ranged from 1.5 to 42 years. Their Knowledge training averaged about two years. The 14.3-year mean describes licensed work, not necessarily the exact experience variable in every correlation.

The comparison database contained 50 healthy, right-handed male controls, scanned on the same system, with a similar age distribution. Female, left-handed, unhealthy or out-of-age-range records were excluded. All 50 were used in the voxel-based analysis. A subset of 16 precisely age-matched controls was used for detailed hippocampal pixel counting against the 16 drivers. These are two control-analysis samples, not two separately recruited groups of taxi drivers.

MRI and two analysis procedures

Participants' brains were examined using structural MRI, with a 2.0-tesla scanner and high-resolution T1-weighted images. Structural MRI describes anatomy; unlike task-based fMRI, it does not show which region became active while a driver selected a route. No navigation task was performed inside this study's scanner.

Two complementary analyses examined the images:

The pixel-counting rater was blind to driver/control status and to the VBM findings. This reduced the risk of drawing boundaries to fit the expected result. The hippocampus was divided along its length into anterior, body and posterior regions. In the reported regional procedure, 24 slices were divided into 6 posterior, 12 body and 6 anterior slices.

The researchers also examined associations with navigation experience, including time learning the Knowledge as well as licensed work. One driver with 42 years' licensed experience was excluded from the experience correlations as an extreme outlier, leaving 15 for that analysis. This does not change the original group-comparison driver sample from 16 to 15.

Using the same scanner and matching age helped reduce measurement and age explanations for group differences. Blinding addressed rater expectations. Neither control removes every difference between an existing occupational group and people outside it.

Regional and experience results

The central finding was regional specificity. Taxi drivers had greater posterior hippocampal measures than controls; controls had greater anterior hippocampal measures. There was no significant group difference in the body or total hippocampal volume. It is therefore inaccurate to replace the result with “taxi drivers have bigger hippocampi” without specifying the region.

VBM detected significantly greater posterior grey-matter measures bilaterally in drivers. The anterior direction favoured controls, but that VBM anterior contrast did not survive correction for the many comparisons. Pixel counting independently confirmed the regional pattern: a posterior advantage in drivers and an anterior advantage in controls, with no larger total. The methods converge on a regional account, while their exact significance details are not identical.

In the experience analysis, longer navigation experience was associated with greater right posterior hippocampal measures and smaller anterior measures. The right-posterior VBM correlation was r = .6, p < .05. The pixel-counting right-posterior correlation was r = .5, p < .06, a weaker reported threshold; that statement alone does not establish significance at 5%. The reported anterior pixel-counting association was negative, r = −.6, p < .05. Do not transfer one method's probability to another method's coefficient.

The experience measure included training, and age was considered in interpreting it. Even after addressing age, the correlations were associations within experienced drivers. They were not pre-training/post-training measurements of the same individuals. Nor did the scan establish that new neurons had grown: macroscopic MRI differences do not directly reveal the microscopic mechanism.

Conclusions, theme and biological area

Maguire and colleagues interpreted the pattern as consistent with experience-related plasticity in the adult human hippocampus. Repeated navigation may be related to the organisation of representations supporting a complex spatial map. The experience association strengthens that interpretation because it relates a regional measure to amount of relevant experience, rather than only to the job title.