This post constitutes the second installment of my trilogy of essays on recent findings of note with functional magnetic resonance imaging. I introduced this series with my post dated Mar. 23, 2009.
In today's post, I discuss a study published online in the journal Nature (2009) on Feb. 18, 2009. Stepheny Harrison and Frank Tong examined where in visual cerebral cortex information may be retained, permitting us to recall visual cues after they disappeared from view. The selection of suitable participants was straight-forward. Alert college students with passable vision sufficed. No confounds were expected from this sampling bias.
The participants were asked to accomplish a simple and robust delayed orientation discrimination task while their brains scanned. That is, sinusoidal gratings consisting of parallel stripes of dark gray with fading edges were shown at two different orientations on a light gray background for fractions of a second. Then a number was flashed on the screen, instructing the participants whether the first or the second grating was to be remembered. After a pause of 11 seconds, a third grating was shown a few degrees rotated against the preceding ones. The participants were required to decide whether this grating was shifted clockwise or counter-clockwise against the orientation of the one they were previously asked to remember. Functional images were acquired for 32 seconds in total.
With this protocol, the participants were exposed to differing information content at constant stimulus intensity, permitting the researchers to detect memory-related changes in local blood flow. As control, the participants were exposed at random to dissociated letters and gratings. In addition, flickering dots were presented at the center of the screen to map the representation of the visual field in cerebral cortex. Much variability in the detection of the stimulus was not anticipated. Thus, the number of participants could be held small. Only six people were needed in this study to obtain statistically significant results. The concepts used in this study are built on a broad body of knowledge on the precise nature of the cortical processing of visual stimuli obtained from research with non-human primates and other animals. This knowledge permitted the investigators to develop a firm working hypothesis for their study. Conversely, their conclusions can be tested in the animal models that provided the basis for the investigators' hypothesis, permitting us to examine the nerve cell mechanisms underlying their findings. This is the strength of the present study.
Gratings can be presented with great temporal accuracy. Nerve cells in the primary visual (V1) area, where visual input feeds into cerebral cortex, respond robustly to this type of stimulus. In addition to V1 cortex, three other occipital lobe areas (V2-V4) are known to contain nerve cells responding to gratings. Nerve cells in areas V1 and V2 are narrowly tuned to specific angles of orientation, that is they become most active when an edge of a particular orientation passes over the part of the visual field that they are sensitive to. However, the nerve cell response ceases within less than a second. One hypothesis posits that sustained nerve cell activity in subsequent processing areas may help retain information about the vanished stimulus.
In order to test this hypothesis, Harrison and Tong compared blood flow changes in areas V1, V2, V3, and V4. Cerebral blood flow increased in these areas within seconds after the onset of the first grating. Though the magnitude varied locally, the differences were not statistically significant. However, the timing and the duration of the change in blood flow between 6 and 10 seconds after the first grating was presented provided a temporal signature with which Harrison and Tong were able to infer the orientation of the grating to be remembered. They were able to identify such orientation-specific temporal signatures in all four visual areas, providing evidence that the memory of peculiarities of transient visual stimuli can be maintained at early stages of information processing in visual cortex.
The elegant simplicity of the design of this study profoundly facilitated the perspicuity of its findings. Based on their observations, Harrison and Tong suggest that local excitatory and inhibitory nerve cell connections may produce sustained, oscillating nerve cell activity, retaining the memory of the orientation of the gratings. This hypothesis can be tested. Recordings of local electrical nerve cell activity from the scalp (EEG) or associated changes in magnetic field strength (MEG) could provide evidence for such activity.
A role for higher order areas in the parietal and frontal lobes known to be involved in memory yet remains to be established. The authors note that compared with the direct response to the gratings, memory-related activation was substantially diminished. As I pointed out in the initial post of this sequel, these areas may well have been activated even less, hidden from examination under threshold.
- Harrison SA, Tong F (2009) Decoding reveals the contents of visual working memory in early visual areas. Nature 458: 632-635.