Experiment 1: fMRI study

Twenty-two participants took part in the fMRI study (average age = 23.81, SD = 4.54; N female = 15). All participants were native English speakers with no history of neurological or psychiatric disorders and normal or corrected-to-normal vision.

There were 4 experimental tasks: episodic, semantic, picture-decision, and control. In each task, the participant was presented with word triads including a target word in the center of the screen and 2 words below, 1 on the left and 1 on the right. The participants had to select the correct option by button press. The words were presented on top of a scrambled or unscrambled picture depending on the condition. The trials lasted 4 s and were preceded by a 1.5-s instruction indicating the upcoming task. The trials were presented using an event-related design with the most efficient ordering of events determined using Optseq (http://www.freesurfer.net/optseq. Null time was intermixed between trials and varied between 0 and 26 s (average = 2.80 s, SD = 3.13) during which a fixation cross was presented. In total, 54 items were presented for each condition. The experiment was split into 3 runs (18 trials per condition), each run lasting 620 s, the order of which was counterbalanced across participants. An example trial from each task can be seen in Fig. 1.

Here the participants were presented with a target word (e.g. knife) and 2 alternative possible features of the object (e.g. sharp vs. bendy), such as its typical function, color, texture, shape, etc. The participants were instructed to determine which alternative was correct. The words were presented on top of a scrambled picture.

Here the word triads were presented on top of a color photograph of an object (e.g. a chair). The target word referred to a property of the picture (e.g. color) and 2 alternative choices (e.g. blue and red). The participants were instructed to select the option that best matched the target feature of the object.

Immediately prior to the scan, the participants were exposed to a selection of 54 color photographs of objects (e.g. a bucket) and told that they would be required to remember aspects of the pictures during the experiment. Each photograph was presented for 10 s and the participants were asked to describe the picture in as much detail to ensure that the pictures were sufficiently encoded. In each trial of the experiment, a target word would be presented (e.g. bucket) and the 2 alternative possible features of the remembered item (e.g. blue or red). The words were presented on top of a scrambled picture. The participants were instructed recall the feature that best described the target item. After a short jittered interval (varying from 0 to 1.5 s), the participants were given 3 s in which to rate the vividness of their memory of that particular item from 1 to 4 (1 = not vivid, 4 = very vivid). The episodic trial and the vividness rating were modeled separately in the general linear model.

In the control task, the word triads consisted of a string of Xs (e.g. xxxxxxxxx) on top of a scrambled picture. The picture was shifted slightly to the left or right. The participants had to indicate the direction of the shift. This acted as control for visual and motor activation.