The notion of phenomenal precision plays an important role in Block’s argument. He (this collection, p. 45 & 47) admits that it is a notion in need of clarification—but one where a lack of definition ought not give us headaches, since many concepts pertaining to phenomenality lack definability.[5]
We can think of precision as connected to bandwidth. What does that mean? Some variations in the external world do not factor into how the world feels to us. For example, one cannot differentiate a grating of 20% contrast from one of 20.2%, or a pain caused by heat of 480 millicalories per second per square centimeter from one caused by 640 mc/sec/cm² (Hardy et al. 1940; Hardy et al. 1952). However, there is a point where the variance in the stimulus becomes just noticeable, e.g., a pain of 660 mc/sec/cm² does feel different from one at 480. This can be measured behaviorally, namely if a subject is able to distinguish one item of type A from another of type B above chance based on the relevant feature (e.g., if 75% of all presented items are distinguished correctly). So all the variance that I cannot distinguish perceptually between two just noticeable differences (JNDs) is covered by percepts with the same phenomenal character. That is, if I have a percept a, different states of affairs may have caused a—and the phenomenal character of the percept does not convey its real cause. So percepts ought to count as a bit imprecise. The more cases are covered by a percept, the less precise: a visual-contrast-percept that can be caused by 20±1% contrast is more precise than one that can be caused by 20±3%.
Percepts have representational and phenomenal aspects—content and character. Precision certainly makes sense when it comes to content, because “[t]he representational content of a perception is—constitutively—the veridicality conditions”, Block writes (this collection, p. 27).[6] So we can look at the range of cases in the world that make a percept veridical, and thereby determine its degree of representational precision based on the range of cases that may have caused it in that obtaining condition. If, for example, a Gabor patch with 22% contrast looks just like one with a 28% contrast, then the representational content of this percept has a degree of precision of at least 6%, because all cases between 22% and 28% are covered by the same phenomenal appearance. Otherwise, these two Gabors would not look the same.
Representational precision makes sense—but how about phenomenal precision? Intuitively, phenomenal precision sounds good: things may appear red or crimson, and because all things crimson are a subset of all things red, the bandwidth of both ways of seeing-as differs—and therefore they ought to count as differently precise.
But if we can diagnose differences in the degree of phenomenal precision, we need a way of estimating its degree. How would we do this? GR provides an easy answer: phenomenal precision is grounded in representational precision, so we can use the same methods by which we estimate representational precision to estimate phenomenal precision. But in an argument where GR is under scrutiny, one cannot presume this without begging the question. So we must look for another way of estimating phenomenal precision.
For this purpose, Block suggests the Phenomenal Precision Principle (PPP), which we may reconstruct as: If the percept of item i1 and the percept of item i2 are phenomenally indistinguishable with respect to some feature F under condition A, but phenomenally determinately different vis-à-vis F under condition B, then the experience in A is less precise than in B.
So if I cannot differentiate two stimuli by their contrast in condition A, but can differentiate the two by contrast in condition B, then my experience in A is less precise than in B. Why? Because if I can tell the two items apart phenomenally, then I can distinguish cases, and therefore the bandwidth of that experience is narrower.
Block uses differences in phenomenal precision prominently in an argument against GR: he believes that in some cases, phenomenal precision (p–precision) and representational precision (r–precision) can fall apart. If GR were true, such that representational features must account for phenomenal ones, then this cannot be the case. But this is exactly what happens, according to Block: “there is evidence that attended and foveal perception can be greater in [phenomenal precision] without involving awareness of more precise environmental properties” (this collection, p. 41). Then, GR is false.