Review Reports - Size Constancy Mechanisms: Empirical Evidence from Touch

Round 1

Reviewer 1 Report

In this article, the authors investigated whether the tactile anisotropy that typically emerges when participants have to estimate the distance between two touches on the hand dorsum is also present when a continuous object touches the skin of the hand dorsum and participants have to estimate its size. The authors present the results of three experiments in which task performance for different shaped objects was compared with task performance for simultaneous touch by two posts. In Experiment 1, with a first object (not curved) results indicated a bias, an anisotropy on the opposite direction that the bias found for the posts condition. In Experiment 2 results indicated there was a bias for the object condition (curved) in the same direction than for the posts condition (though smaller than for the posts condition). This smaller bias for two different object conditions (both curved, one of them with constant surface size across trials) in the same direction than for the posts condition was replicated in Experiment 3 and in this case the bias was not significantly different from a non-bias scenario.

The paper is well written, and it is relevant for the field of multisensory research. The introduction makes a good work setting the scenario and providing references to related work, the experiments are beautifully performed, and there is a good discussion on the possible mechanisms behind the results. I have two bigger points which I hope the authors can clarify in their revision, and a number of minor points, which I summarize below.

Motivation for the study: while the Discussion makes a great work showing the relevance of the research question in terms of learning about mechanisms underlying body and tactile perceptions, this relevance is not that clear in the Abstract or Introduction. I suggest to authors to explain in the abstract the relevance of learning about how people estimate the size of a continuous object. Similarly in the Introduction the paragraph in Lines 111-123 needs a better justification on the advance/research question that the new study would bring with respect to the one by Anema et al. 2008. Since Anema et al replicated the Weber’s illusion with a solid object, why would one expect different results in terms of anisotropy on the dorsum of the hand when object size vs distance between two touches has to be estimated? This is not clear.

Results Interpretation: I have trouble with this general conclusion in Lines 413-415: “Critically, however, we did not find this bias when the stimuli were continuous objects (Experiments 1-3). Moreover, the absence of this bias with continuous objects was not due to the curvature of the objects”. In Experiment 1, with the first object (not curved) results indicated a bias, an anisotropy on the opposite direction that the bias found for the posts condition; and in Experiment 2 results indicated there was a bias for the object condition (curved) in the same direction than for the posts condition, (though smaller than for the posts condition) as the PSEs were significantly lower than 1. Thus, results for the curved object in Experiment 2 do, to some extent, show a bias in the same direction (though smaller) to the one found for the posts, and the curvature of the object does have an influence in the bias. These General conclusions, should be rephrased to reflect these results. Similarly, the sentence in Lines 426-427 (“In the case of objects, we propose that such a mechanism is simply more effective, given that anisotropy was absent or negligible.”) and the Abstract should be toned down, given the significant results obtained in Experiment 2 that indicated there was a bias for the object condition (curved) in the same direction than for the posts condition (though smaller than for the posts condition).

Below are some other minor points that I would advice the authors to address in a revised version of the paper:

Introduction:

Line 68: having defined before size constancy for vision as “a cognitive mechanism that allow us to perceive an object as having the same size even when seen at different distances” it is not clear at this point to what size constancy refers to for a sensory modality like touch, when objects are only perceived when being in contact with the skin (or very proximal). Could you add a definition?

Experiment 1

Lines 185-187: Procedure: was the order of experimental block (i.e., posts vs object) counterbalanced across participants?

Experiment 2

Line 264: Participants: had any of the participants taken part in Experiment 1?

Experiment 3

Line 340: Participants: had any of the participants taken part in Experiments 1 or 2?

Line 378: Hand shape: I am lost here, how was “hand shape” calculated? This is the first mention of it.

Lines 374-376: This part requires further explanation in Discussion: In Experiments 1 and 2 it was found that “the mean slope for posts was significantly greater compared to objects”, however in Experiment 3 this was not replicated as the mean slope for posts was significantly greater compared to same surface objects, but not for different surface objects”.

General Discussion

Lines 541-545: please split this sentence into two sentences, as it is very complex.

Figure 1: I would specify in the figure legend that A was used in all experiments, B in Experiment 1, and C in Experiments 2 and 3

Figure 5: which of the two types of objects is represented for Experiment 3?

Typos:

Line 272: “the experimenter were”

Line 328: “to the test” -> to test

Line 332: “when object’s length was increases the width…”

Line 484: “an object that touch” -> “an object that touches”

Author Response

Thank you for the positive notes on our work and your constructive comments. A point-by-point response to the reviewer’s comments is enclosed below and a revised version of the manuscript has been submitted. All changes have been highlighted in the manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper makes a very nice empirical contribution, via three experiments, to the data on two-point touch anisotropies. To be upfront, I’m a vision researcher, not a touch researcher, so I read this article as a generalist reader, not as a tactile specialist.

The starting point for the paper is the well-established finding that two-point touch distance perception is anisotropic. Specifically, distances along the hand’s length are perceived as longer than equal distances across the hand’s width. But in the intro we are not told much about why that might be. Instead, we move on to the question of whether the same anisotropy will hold if the distance is signaled by a continuous object rather than by the two points at its end. That’s a good empirical question, motivated by the concern for ecological validity. That is, we encounter whole objects much more frequently in life than two-point touches. But a puzzle that remains for me is how is the question of whole object contact related to the question of two-point touch with regard to size constancy processes? From what I can surmise, it seems the authors think the two-point anisotropy should be even larger than it is, based on some analysis of receptive fields that is not fully explained in the intro. The reason it is not is attributed to an unspecified size constancy process. And I think — though I’m not 100% sure — that they suspect the size constancy process will be able to fully compensate for the anisotropy when a whole object is used.

The methods are described very clearly. All seems in order there. Design decisions are properly rationalized and explained. The data are also very clear. They would make an excellent model for teaching purposes about psychophysical functions, points of subjective equality, and just noticeable differences.

The findings seem pretty solid. The two-point anisotropy is replicated and stable in 3 independent experiments. The anisotropy is much reduced (sometimes even gone) for whole objects. The three experiments differ only in subtle ways in which the whole object condition is implemented, first with a straight edged surface, then with a gently curved one, and then one with a constant surface area (thinner to compensate for longer).

The whole object condition shows two main differences from the two-point condition. It varies a little with every change in implementation and it is noisier (the just noticeable difference is always larger). The authors have some reasonable ideas about why the little changes occur across experiments. They leave the question of the greater noise to a final short paragraph, where they offer one reasonable idea (i.e., the double touch in the center of the hand when whole objects are compared). Not very convincing in my view, but I realize it is also a tangent from the main thrust.

Title. It seems a little odd to have vision in the title when the study presented is entirely about touch. I recommend limiting the title to “…touch constancy.”

Intro. The beginning of the paper has quite a slow windup. I recommend the authors signal the main contribution of the paper much earlier, preferably in the first paragraph. Doing that will help guide the reader through the material that is now presented in the windup.

We are also told very little about the presumed basis for this anisotropy in the introduction. Instead, we are told as an aside, that its origin “…has been proposed to derive from a process of tactile size constancy that compensates for the distortions only incompletely (Longo, 2017a).” For the generalist reader, that’s not enough. The reader should not have to read the 2017 paper to move forward. A quick summary of the main idea is in order. I realize that much more detail is given in the discussion, but that’s also too far removed to assist the reader here. So a missing piece for me in the intro is how does some size constancy process account for the anisotropy typically found in two-point distance perception on the hand?

The predictions leading into the method left me a little lost. The authors say, “If the same mechanism of tactile size constancy applies both on distance and size estimation of tactile stimuli we should have the same level of anisotropy for both distance and size estimation. Instead, if two different mechanisms mediate the tactile distance and size estimation anisotropy should vary depending of the type of stimulation - i.e., points or object.” Are we talking about TWO mechanisms of size constancy here? Or the same size constancy process that is less or more complete, as implied earlier in the intro? Perhaps some guidance about how size constancy (whether only partial or complete) might be accomplished is required in the intro already.

Results. “Meta-analysis” is not the correct label for a straightforward statistical comparison of the data in the three experiments.

Discussion. It rambles. I was admittedly lost in several places. I think I could do a reasonable job of presenting the main ideas, but I would not be confident. There is an appeal to lateral inhibitory mechanisms to account for differences in apparent length between two-point and whole object conditions. But I can’t tell how that idea is related to the larger anisotropy for two-point than whole object conditions. Perhaps another panel in the figure should be added to help visualize that?

Overall. The data are great; a really sold contribution. The writing is wordy and often rambling. The introduction and discussion need to both be rethought and maybe reorganized to lead general readers like me in a straighter line from two-point anisotropy to size constancy (via feedback or lateral inhibitory mechanisms?), and then from there to why size constancy might be more complete for whole objects than for two-points.

Author Response

Author Response File: Author Response.pdf