Temporal coding in neural populations
Single-unit recordings reveal compressed sequential activity and slowly varying temporal context signals that can support a mental timeline.
Does the human mind organize the recent past and the near future along a common ordered mental timeline? And if so, what are the properties of that timeline, and how do we navigate it?
We study that question with carefully controlled behavioral tasks that isolate temporal structure from meaning, then use response-time patterns to infer how that internal representation is searched.
Specifically, we used two tasks: judgment of recency and judgment of imminence. In judgment of recency, participants see a short sequence of letters and later choose which of two probes appeared more recently. In judgment of imminence, participants first learn a probabilistic sequence and then choose which of two probes is expected to occur sooner if that sequence continues. The two tasks are designed to ask parallel questions about memory and prediction.
Across both domains, the data support the same idea: people scan along an ordered internal representation whose resolution is finest near the present and gradually coarser farther away, yielding a compressed timeline for both past and future.
The key diagnostic signal is the change in response time with probe position. In judgment of recency, correct response times depend strongly on the lag to the more recent probe and not on the lag to the less recent alternative. In judgment of imminence, the same asymmetry appears for the more imminent probe. That is the pattern expected if people start near the present and scan until they encounter the correct item.
Importantly, response time does not rise linearly with lag. Instead, it increases sublinearly, which is what we would expect if the representation is compressed: nearby events are spaced farther apart in representational time than distant events, so scanning slows most near the present and becomes progressively coarser farther away. The rate of compression is well described by a logarithmic function, as expected under the Weber-Fechner law.
Try a small judgment of recency experiment
Letters appear one at a time, as they do in the laboratory task. After each list, answer from the keyboard: F or Left Arrow for the left probe, J or Right Arrow for the right probe, and K or Up Arrow if you are unsure. The demo records response time on every trial and updates the lag plots continuously as you go.
0 trials completed
+
F or ←
J or →
Unsure: K or ↑
Accuracy by more recent lag
Accuracy by less recent lag
Correct-response time by more recent lag
Correct-response time by less recent lag
Accuracy includes all trials so far. Response-time plots average only correct left/right responses.