No problem.

We need to be careful to differentiate between changes in prices and expectations of these changes. If the expectations are mostly accurate, the direction of the change is irrelevant. The only reason why people might consume more during an inflation is if they are tricked, i.e. if they think the prices will rise slower than they really will.

There are several issues here too. First of all, the price of the capital goods is derived from the price of the consumer goods. Unless there is friction in price adjustments then, it's impossible for the unit of account to have a negative nominal interest rate. The real interest rate can't be negative even irrespective of that, as that would mean that people value the same good in the future more than they value it in the present, and that contradicts the axiom of human action.

Of course, a negative nominal interest rate can hypothetically occur with something that is not a unit of account. But this is not as big a problem as mainstream economists assume. I did a simulation and presented the results at the London Bitcoin Conference 2012. In my simulation, a merchant with a 6% real markup and a turnover/stock ratio of 6 was able to have a positive nominal RoI using a deflationary currency that is not unit of account even at a 50% deflation (i.e. the prices in that currency halve within a year). Even income taxes did not revert the positive RoI (of course that depends on the details of the tax rules, but I took those that according to my research should be applied to Bitcoin in Europe).

In fact, because the prices dropped, after a year the simulated merchant would be able to double his stock, i.e. he saved enough capital to be able to expand his business. This is a hint that a deflationary currency edit: falling price level is highly beneficial for consumers, as it increases the amount of goods available to them. And if you don't believe the simulations, just look at consumer electronics.