Suppose all of log2 fold changes from 1000 genes are negative (down regulated)

library(gage)

go.set <- go.gsets(species = "human")
go.cc.hs <-go.set$go.sets[go.set$go.subs$CC]
fcs<- -c(1:1000)
set.seeds(123)
names(fcs)<-sample(unlist(go.cc.hs), 1000) 
head(fcs)
  7392  10139  26470  91039 170575    214 
    -1     -2     -3     -4     -5     -6 
 x<-gage(fcs, go.cc.hs)
head(x[["greater"]])
                                                  p.geomean stat.mean
GO:0035770 ribonucleoprotein granule             0.01073012  2.423047
GO:0036464 cytoplasmic ribonucleoprotein granule 0.01816832  2.195199
GO:0005770 late endosome                         0.07821994  1.437179
GO:0044440 endosomal part                        0.08148658  1.404548
GO:0005844 polysome                              0.08386275  1.439267
GO:0019867 outer membrane                        0.08875160  1.386214
                                                      p.val     q.val set.size
GO:0035770 ribonucleoprotein granule             0.01073012 0.7266728       17
GO:0036464 cytoplasmic ribonucleoprotein granule 0.01816832 0.7266728       16
GO:0005770 late endosome                         0.07821994 0.7266728       28
GO:0044440 endosomal part                        0.08148658 0.7266728       56
GO:0005844 polysome                              0.08386275 0.7266728       10
GO:0019867 outer membrane                        0.08875160 0.7266728       14
                                                       exp1
GO:0035770 ribonucleoprotein granule             0.01073012
GO:0036464 cytoplasmic ribonucleoprotein granule 0.01816832
GO:0005770 late endosome                         0.07821994
GO:0044440 endosomal part                        0.08148658
GO:0005844 polysome                              0.08386275
GO:0019867 outer membrane                        0.08875160

Since all fold changes are negative, I am wondering how the "greater" pathways have been calculated?