OnMouse down does not work with a mesh collider

I am working on a project in which I want to select and drag complex objects with the mouse. I have previously done this in another project using regular colliders and the OnMouseDown function, but this is not working with mesh colliders.
For now, I have a simple script that is supposed to change a public bool when the object is clicked ( I am using OnMouseDown). I am not detecting any click when I click on the complex objects with mesh colliders. I have tried using the same script on the objects with a sphere collider and also on a cube with a mesh collider. Both of these worked, so I do not think there is a problem with my script.
These objects were downloaded online and I am not sure how they were made. I imported them into Blender before importing them to unity after.
Can anyone think of why my script is not working or do you have any ideas for how I could select and drag the objects in a different way? Thank You

It depends on what you are going to do with the brain model in Unity but realistically I’d say it’s a bit too complex for most game uses anyway. Or at least unnecessarily complex. Each hemisphere of the brain is over 38,222 tris. All parts combined are 94,716. That’s significantly higher than characters, objects, buildings etc use in modern games.

If the brain is the sole focus of the game (like an anatomical lesson on the brain) then that might be ok, although as the whole model only breaks into 6 parts it probably isn’t detailed enough for that. If you intend to have other things in the game or multiple instances of this brain then it is really far too high. Either way the tessellation could definitely be heavily optimised and a lot of the ridges and furrows created with normal maps instead.

However I also see a good deal of pinches in the mesh, weird normals, some gaps in the mesh and edges that intersect faces so in all honesty optimising it would probably take longer than it would be worth compared to finding an alternative model which is lower poly to start with. If you do use this model be aware that these errors may be visible from some angles and make it look a little ugly.

If you just need to detect clicks on the brain parts without them colliding with anything else then they can use a non-convex MeshCollider however they cannot have a rigidbody on them then. Still 38,222 tris is far too high for a collision mesh though and it would need to be cut down significantly for performance reasons.

If you need the parts to actually collide they would have to be convex and that means the collision mesh would need to be below 255 tris (not sure why the error message says polygons when the documentation says triangles).

To create a collision mesh for this you basically just want to model a mesh that describes the rough shape of the brain parts with as few triangles as possible. Even running this model through the automatic optimiser in Max at 1% vertex count it’s still too high for a convex mesh though. All of those errors I mentioned also cause problems with this so however you do it a degree of manual work is going to be required.

You can probably find some good tutorials for creating collisions meshes in Blender online.