How to plot predicted values for fixed and/or interaction effects? (different results of predict() function and effect() (from effects package))

Question

I've applied a linear mixed model to my data:

g1rfinal = lmer(log(trmLatency) ~ WMC + order + distance + DomSide + (1 + BlockType | 
pno) + (1 + firstBlType | pno) + BlockType + (1 + distance | pno) + WMC:order + WMC:distance + WMC:DomSide + order:DomSide + 
distance:DomSide + WMC:BlockType + distance:BlockType + DomSide:BlockType + 
order:BlockType, data = g1Study)

"order" variable is a continuous predictor and "BlockType" is a factor with two levels. I want to plot predicted values to check the effects of "order: BlockType". (I've checked the effect with chi square test and it was significant). I've got two ways. One of them is predict() function. I added predicted values to my dataframe and averaged them to be grouped by order and BlockType:

g1Study = cbind(g1Study, predict = exp(predict(g1rfinal)))
g1test = g1Study %>% group_by(BlockType, order) %>% summarise(avgrt = mean(predict))

Then I plotted the values from g1test and the values from effects package:

ggplot() + geom_point(data = g1test, aes(x = order, y = avgrt, color = BlockType)) + geom_line(data = as.data.frame(effect("order : BlockType", g1rfinal)), aes(x = order, y =exp(fit), color = BlockType))

What I've done for averaging predicted values seem to be identical for what effect() function does. effect package help page

But I get different results: I can't get what's happening here.