36 THE SCIENTIST | the-scientist.com
the beginning of trials, and then that activity decayed as individual
neurons relaxed at various rates. Other cells in the LEC, meanwhile,
seemed to become gradually less (or sometimes more) active over the
course of the entire experiment. Looking at the data this w ay, the team
was able to distinguish individual trials not just according to wall color
but, far more intriguingly, by the order in which the rat had done them,
explains May-Britt Moser. “Together, [these cells] coded for time.”
Publishing the findings in late 2018, the team cited Howard’s and
Shankar’s work, highlighting how the sort of activity patterns Tsao had
seen in the LEC neuronal population matched up with the pair’s theo-
retical predictions.^7 The Norwegian group also noted that this evolving
signal seemed able to track passing time over multiple timescales—
changing fast enough to distinguish between individual moments on
the scale of seconds within a single episode, as well as to distinguish
whole episodes from one another over the scale of minutes or hours.
On reading the team’s findings, “I was ecstatic,” Howard says. “It was
really a big deal for me.”
The paper was exciting for many in the neuroscience community,
and its publication was followed by a burst of theoretical work from
several groups, not just Howard’s. Edmund Rolls, a computational
neuroscientist at the University of Warwick, incorporated the find-
ings from the Kavli group’s 2018 paper into a model that explored
how interacting networks in the brain might convert gradually
changing LEC activity into a sequence of hippocampal time cells,^8
based on a framework he’d developed more than a decade earlier
to explain how grid cells might lead to the generation of place cells.Additional experimental data started flowing in, too. Howard
and colleagues, for example, analyzed recordings from monkeys’
entorhinal cortex—an area containing the MEC and LEC—and
found activity similar to that observed in Tsao’s rats, according
to a preprint published last summer on bioRxiv. Specifically, a
cluster of neurons in the entorhinal cortex spiked after a mon-
key was presented with an image, and then returned to baseline,
with different neurons relaxing at different rates.^9 Just a couple
of months later, researchers in Germany reported that activity
recorded from the human LEC could be used to reconstruct the
timeline of events people experienced during a learning task.^10
The gradual change in LEC activity wasn’t the only novel result
from Tsao’s paper. Several groups picked up on a related finding
that the rate of change in the LEC—and indeed in many areas of
the brain—may depend on the sort of experience an animal is hav-
ing. That phenomenon might help explain why the passage of time
within episodic memories seems so subjective.Personal time
As a follow-up to his original experiments with the rat arena, Tsao
had done a couple of additional trials during his Kavli intern-ship with a figure-eight maze. In each of those trials, instead of
freely exploring an arena, the rat would run around the maze,
following the track left, then right, then left, and so on. After dis-
covering patterns in rats’ LEC neuronal firing during arena tri-
als, Tsao hoped to see something similar in data from the figure-
eight mazes—something that would distinguish trials from one
another according to when they took place. “But... it turned out
we couldn’t tell them apart very well,” says Tsao. “For a while this
was very disappointing—this was basically the opposite conclu-
sion that we had reached from the [arena] experiment.”
It wasn’t until Tsao dug into the literature on episodic memory
that he came to realize what might be going on. “Maybe it’s not so
much about physical time, as you measure in clocks, but more about
subjective time, as you perceive it,” he says. Running in a twisted loop
was a repetitive, boring task compared to exploring an arena, and
the rat’s LEC seemed to reflect that by changing its activity less sub-
stantially during the figure-eight experiment than it had during the
arena experiment. It seemed as though the rat’s brain wasn’t really
experiencing individual figure-eight trials as distinct events, at least
not to the extent it had for arena trials, Tsao says.
This link between the type of experience and the way time is
represented in neurons touches on a well-known quirk of episodic
memory. It’s easier to pick out memories from a week of exciting and
varied activities than from a week filled with normal, uninterest-
ing tasks, and the former feels much longer than the latter when it’s
recalled. (This is different from the sensation of time dragging when
doing something boring—an effect of consciously counting time as
it passes rather than representing it in a memory of the event, notes
Buonomano.) Tsao’s study hinted that part of this subjective effect
might arise because the LEC, which receives neural input from areas
involved in processing sensory information, changes its activity to a
greater degree during more complex experiences than during ones
that require little processing. It implies, Tsao speculates, that time in
memory might be entirely “drawn from the content of your experi-
ences, as opposed to being coded as an explicit thing.”
Although neural recordings are challenging to carry out in
humans, functional MRI (fMRI) data from other research groups has
helped flesh out the link between the rate of activity changes in the
cortex and the representation of time in memory. Kareem Zaghloul,
a neurosurgeon and neuroscientist at the National Institute of Neu-
rological Disorders and Stroke and a “big fan of Marc Howard’s work
and his model,” had been running an experiment on the effects of
brain stimulation on human memory around the time Tsao’s paper
came out. As part of their project, Zaghloul and his colleagues decided
to use their dataset to look at how temporal context might influence
memory formation. “We hypothesized that perhaps the extent to
which these signals of time change, maybe that affects your ability to
distinguish memories from one another,” Zaghloul says.
Participants in his group’s study had been asked to learn pairs of
words, such as “pencil” and “barn,” and then remember these pairs
later while avoiding confusing them with other pairs they’d learned,
such as “orange” and “horse.” Measuring activity using electro-
encephalography across broad regions of participants’ brains whileTogether, these cells coded for time.
—May-Britt Moser, Kavli Insti tute