"Opinionated jerk" or not, Montello might be selling the cultural anthropologists short, particularly by dismissing, in whole, the work of Ed Hutchins in distributed and embodied cognition. Whether spatial cognition is or is not ultimately culturally dependent; it is most certainly embodied, with spatial referents and means of reference (language, gestures, artifacts) changing depending on the context. Thus the boundary for downtown as constructed from Montello’s map drawing task would almost certainly be different than the boundary as constructed from another method. What would have happened if he had asked his subjects to identify downtown by its landmarks such as buildings or roads? What would have happened if he had asked them to walk to the “edge” of downtown or to walk around the “edge” or downtown? (Besides, the subjects refusing to participate.) What would have happened if he had presented a different kind of map or perhaps no map at all? Granted that taken in aggregate these differences might not be noticeable in a point map, but to the subject acting then and there, they are not trivial. And for any organizations concerned, if I subjectively or we collectively feel more or less in downtown given the situation, then THAT may be something to be interested in.

I think I was unclear, in the "downtown" research case, as to the specific goals Montello's research hoped to achieve. If it was to simply support GIS, then I'd wager there are more appropriate ways to address the question. For instance, assuming city planners are not completely "off-base" when they formally designate a part of town as "downtown", perhaps we could simply aggregate all the formally specified downtowns (in North America, say), attach some features to them, and then run them through an auto-encoder to "average out" the city-specific differences. GIS and geography in general are nothing if not data-rich, and it seems a shame to not make use of that.

At some level, I think the big picture goal of Montello's work is to provide evidence for the need to consider how people think about geography when designing GIS. While a region might have a particular official designation, usage often differs. Consideration of these differences is very important to the creation of usable GIS. That said, it is unfortunate that he does not delve deeper into some of the the issues. He acknowledges that there are context effects in terms of how someone may define something like "downtown", but doesn't really say anything about what these are and how they should be considered. For example, I imagine that some regions may have definitions that have changed over time, or boundaries that were defined for historical reason. The amount of time someone has spent in an area might influence how they define the boundaries. In the "downtown" paper, it is mentioned that they recorded the amount of time each person lived in the area. It would have been interesting to use this not just as a cut-off, but as an additional point for analysis.

I found Montello's talk and papers interesting in a several ways. I thought his idea of talking about regionalization as a kind of categorization was interesting bacause I never thought about this phenomenon in this way. I was hoping that he would spend more time talking about how context affects spatial orientation - for example, what would happen if the researchers in the "downtown" study would collect the data in different parts of Santa Barbara?

In my opinion, the greatest value of Dr. Montello's work is that it engenders respect for the subjective aspects of geography. Although we've come to expect that defining place is a top-down enterprise administered by those in power, bottom-up personalized geographies play out in many ways in every day life. At a very subjective level, Andrea Chiba said that we encode our personal memories with a "place stamp."
Regarding Montello's study about spatial relationships between North American cities, it is worth mentioning that when people mistake Reno as being to the east of San Diego they aren't stupid. (Well, they could be, but that's another issue.) They are applying a rule of supra-ordinate relationships that has probabalistic advantages.

I thought his talk was interesting overall but it did contain some unclear points. It was interesting because it was trying to give his audience a broader understanding of geography research. It would have been more interesting if he had more experiments for the "downtown" research in other areas or experiments of similar type for areas other than downtown. His points about north american cities were not very convincing to me.

I have some questions about the "category/region adjustment model" of location estimates Montello discussed in his talk. Are both the coarse- and fine-grained representations proposed by this model in an allocentric format? Why do we carve up spaces into regions, and how do we go about doing it? Is it simply a heuristic to efficiently track locations in space? Is either component in any way tied to our embodiment and interaction with things in the world, and thus perhaps making use of egocentric spatial representations in a substantive way (i.e., actually used for processing, rather than simply perceptual representations being in egocentric formats and needing to be translated to an allocentric format for further processing of the kind proposed by the mode)?

As someone who is very bad at geography, I agree with Nancy's point. In the same way that Leo suggests making generalizations about what constitutes "downtown" by relying on a mass of data, it's reasonable to think of Nevada as being east of California, and to base geographical judgments on that knowledge (San Diego is in California, Reno is in Nevada, and Nevada is east of California; therefore, Reno is east of San Diego). By "reasonable", I mean that this strategy works well most of the time, giving rise to the "probabilistic advantages" Nancy mentions. (I was going to cite other examples of bad generalizations I make, but then I realized other people might not share them...)