THEORY-1289: In a dynamical model anchored to an identified mechanism, holding one part's variable fixed and finding that the phenomenon persists indicates that the part is not a component of the mechanism responsible for that phenomenon, as when Smolen et al. fixed CLOCK and circadian oscillation continued; the verdict is relative to the phenomenon and only as good as the model, and it decides which pathways the phenomenon needs, not which lie inside the system

Source

Bechtel and Abrahamsen (2010), LIT-1936, §3.3, p. 326, reporting Smolen, Baxter and Byrne (2001); read in NOTE-1492.

What was actually shown

• The question. The Drosophila clock has a negative loop, PER:TIM inhibiting CLOCK:CYCLE’s activation of per and tim, and a positive loop, CLOCK promoting its own transcription. Does the positive loop “cancel out” the negative, or is it needed for oscillation (p. 326)?
• The test. Smolen et al. built a reduced model. PER and TIM were merged into one variable, and transport was replaced by a delay. It oscillated with a period of 23.6 h, robust to parameter changes. They then “fixed the value for CLOCK concentrations in the model. The model continued to generate oscillations”, still robustly.
• The reading (C3). “A model can indicate which parts and operations of a system that have been identified empirically are essential for producing the phenomenon of interest, and therefore should count as components of the responsible mechanism.” A reduced model makes it easier to see “which are components of the system but not of the responsible mechanism for that particular phenomenon”.
• Why in a model. The same manipulation in vivo would need “a buffer, or otherwise stabilizing the concentration of CLOCK”; “such exploration is much easier to accomplish in a model”.
• What could have failed. Oscillation could have stopped when CLOCK was clamped, as it would if the positive loop were needed.

What this does not say

• That CLOCK plays no role in the clock. It is the transcriptional activator. What the clamp removes is the oscillation of CLOCK, the positive loop’s contribution, not CLOCK itself.
• That the verdict holds in the fly. The model omits and merges parts. The paper’s general caution applies: model experiments “do not show that the mechanism in fact could produce these effects” (p. 325).
• That a part found necessary is inside the system. The clamp tests whether the phenomenon needs a pathway. A pathway can be needed and located outside the boundary of the system said to have the phenomenon, as McClelland et al.’s externally supplied hippocampal training is needed for their model’s consolidation (p. 331). Where the boundary runs is a separate input (THEORY-1287, CLAIM-530).

Connections

• The record’s knockouts. “Removing the c₂ → c₁ production collapses both capacities” (CASE-189). “Removing any single production pathway sends all three to zero” (CASE-189, Construction 2). The knockout in CASE-190 is the same. Each is a C3 test in a model. Each shows that the maintenance phenomenon needs that pathway. Since the models stipulate which terms are internal, each shows nothing about whether the pathway is the organization’s own.
• CLAIM-516. Removing a feedback edge set extinguishes every capacity. That is C3 applied to every cycle at once, and proved rather than simulated.
• CLAIM-414 and CLAIM-522. The separating intervention in each is a clamp or knockout of a production pathway. This paper’s C3 is the same move in a model of real biology.