What Are We Investigating?

Researchers are delving into the origins of thought. Sometimes, it feels as if we generate our thoughts independently: “this is my idea.” Conversely, in cases of post-traumatic stress disorder (PTSD), intrusive thoughts can arise involuntarily. Individuals with PTSD often report a lack of control over their disturbing mental imagery.

Previous studies have shown that the imagery a person visualizes is reflected in neural signals within the visual cortex. Additionally, it has been established that neural signals in the prefrontal cortex can predict decision-making. By combining these insights, researchers aim to determine whether we can predict which of two images a subject will visualize based on their neural signals.

Section 1.1: Research Methodology

To explore this further, scientists devised a unique mental imagery task for subjects in an fMRI scanner. Initially, participants view images of two contrasting grating patterns: horizontal green/vertical red bars or horizontal red/vertical green bars. The fMRI data collected is then processed through a machine-learning algorithm designed to decode the neural activity associated with each grating.

Now equipped with the neural firing patterns corresponding to each grating (consider them as templates), researchers can compare real-time fMRI activity against these templates to ascertain which grating the participant is visualizing. A critical aspect of this research is measuring the time lag between the activation of neural patterns linked to one grating and the moment a participant indicates they are imagining it. Essentially, they seek to identify the gap between predictive neural firing and the conscious engagement with the image.

Participants are allowed to choose which grating to imagine and when, enabling researchers to differentiate between predictive signals and those that arise during active imagination. Each trial begins with a “pre-imagery” phase lasting up to 20 seconds, where subjects focus on a fixation point on the screen while deciding on a grating. Following their selection, they initiate a 10-second “imagery period” to vividly visualize their chosen grating. Afterward, participants answer two questions: which grating they imagined and the vividness of their experience.

Section 1.2: Key Findings

The results yielded fascinating insights. Activity patterns in frontal and visual regions, among others, were found to predict imagery content as early as 11 seconds before participants decided which grating to visualize. These neural patterns resemble those that encode sensory information.

The brain network responsible for encoding imagery before conscious thought involves four key regions: the frontal lobe (which governs executive functions), the occipital lobe (the visual cortex), the thalamus (which relays sensory information), and the pons (which manages sensations and activities such as breathing). It is logical that these areas would harbor imagery-related information prior to the imagery phase.

To ensure the reliability of their 11-second measurement, researchers conducted various control experiments. They tested their decoding algorithm to confirm it accurately classified imagery information across all trials. This process can be complex due to multiple variables (e.g., potential "spill-over" neural activity between trials).

Additionally, researchers explored whether participants began imagining the grating before the official imagery period. They employed a method known as “binocular rivalry” to evaluate the precision of thought reporting. The outcomes indicated that subjects accurately reported their imagery timings during this control experiment, although it was conducted outside the fMRI environment.

A final noteworthy discovery was that the decoder performed more accurately when participants rated their imagery vividness higher at the trial's conclusion. This suggests that the neural networks responsible for predicting which image a subject will visualize also correlate with the intensity of that visualization.